The2019COMBINEMEETINGco.mbine.org/sites/default/files/COMBINE_2019_Book.pdf · Predictive in-silico...
Transcript of The2019COMBINEMEETINGco.mbine.org/sites/default/files/COMBINE_2019_Book.pdf · Predictive in-silico...
The 2019 COMBINE MEETING
Heidelberg15-19 July 2019
ii
Welcome to Heidelberg!
In 2010, the first COMBINE Forum was held in Edinburgh (UK). Since then, themeeting took place as the annual meeting of the whole COMBINE community.We are very proud to host the tenth COMBINE Forum in Heidelberg this year atthe Heidelberg Institute for Theoretical Studies (HITS) from July 15th to July 19th,2019. More than 100 guests from around the globe will celebrate with us, exchangingtheir ideas, approaches and implementations for improved data and modelling stan-dards in the life sciences. We will as well discuss further needs for standardizationand harmonization in order to further improve data and model interoperability. Acombination of keynote lectures, invited talks and interactive breakout discussions,as well as contributed talks, posters and lightning talks, selected from submittedabstracts, provides the basis for the meeting, offering diverse formats to work oninteroperability problems.
One special focus this year is on the standardization need in systems medicine,which has been recognized as a necessity for computer-assisted personalized medicine.To direct the scientific communitys attention to this topic, the European stan-dardization framework for data integration and data-driven in silico models (EU-STANDS4PM) organizes a workshop as part of COMBINE 2019.
Also, reproducibility in modelling is a main topic this year, as reflected by severalsessions and workshops, like the FAIRDOM PALs and user workshop, the inauguralmeeting of the Model eXchange consortium and breakout sessions on data and modelreproducibility, as well as on metadata specifications. Besides these focus themes,also sessions, workshops and breakout discussions around single standards, theirfurther development and their interoperability help to advance the standardization,standing in the tradition of COMBINE.
We invite you to celebrate the tenth anniversary of the COMBINE Forum thisyear with us and wish you a productive meeting, fruitful discussions and a wonderfultime in Heidelberg.
Martin Golebiewski & Dagmar WaltemathConference Chairs(on behalf of the organizing team of COMBINE 2019)
iii
Funding is gratefully provided by:
iv
Welcome by the HITS management
On behalf of the HITS management, we are pleased to welcome all delegates andvisitors to the campus for the 10th COMBINE meeting which runs from 15-19 Julyat the Studio Villa Bosch, right next to the Heidelberg Institute for TheoreticalStudies (HITS).
There is some history of COMBINE at HITS. After our institute was foundedby Klaus Tschira and Andreas Reuter in 2010, the following year already saw thehosting of the second meeting at HITS. It was a major effort for the young instituteas it was then, since COMBINE has been the combined conference on standards incomputational modelling of biological systems.
The COMBINE community has returned to HITS stronger than ever for its10th anniversary meeting.
We are delighted that we will be host to this event again and we hope you willjoin us to make this anniversary meeting a memorable event.
Dr. Gesa Schonberger (HITS Managing Director) &Priv.-Doz. Dr. Wolfgang Muller (HITS Scientific Director)
About HITS
The Heidelberg Institute for Theoretical Studies (HITS) was established in 2010 bythe physicist and SAP co-founder Klaus Tschira (1940-2015) and the Klaus TschiraFoundation as a private, non-profit research institute. HITS conducts basic re-search in the natural sciences, mathematics and computer science, with a focus onthe processing, structuring, and analyzing of large amounts of complex data and thedevelopment of computational methods and software. The research fields range frommolecular biology to astrophysics. The shareholders of HITS are the HITS-Stiftung,which is a subsidiary of the Klaus Tschira Foundation, Heidelberg University andthe Karlsruhe Institute of Technology (KIT). HITS also cooperates with other uni-versities and research institutes and with industrial partners. The base funding ofHITS is provided by the HITS Stiftung with funds received from the Klaus TschiraFoundation. The primary external funding agencies are the Federal Ministry of Ed-ucation and Research (BMBF), the German Research Foundation (DFG), and theEuropean Union.
v
About COMBINE
Standards for exchange of scientific data are critical to the development of any field.They enable researchers and practitioners to transport information reliably, to applya variety of tools to their problems, and to reproduce scientific results. Over the pasttwo decades, a range of standards have been developed to facilitate the exchangeand reuse of information in the domain of representation and modeling of biologicalsystems, especially in the fields of systems biology and systems medicine. Thesestandards are complementary, so the interactions between their developers increasedover time. By the end of the last decade, the research community decided that moreinteroperability is required between the standards, and that common developmentis needed to make better use of effort, time, and money devoted to these activities.
As a consequence, the COmputational MOdeling in Biology NEtwork (COM-BINE) was created beginning of 2010 to enable the sharing of resources, tools, andother infrastructure, and to coordinate standardization efforts for modeling in biol-ogy. COMBINE began with four core standards, and it has since expanded to eight,and coordinates with a number of other related standardization efforts. COMBINEbrings standard communities together around activities that are mutually beneficial.These activities include making specification documents available from a commonlocation, providing a central point of contact, and organizing regular face-to-facemeetings.
To this end, the COMBINE network organizes an annual conference style meet-ing (the COMBINE Forum) and annual hackathon style events called HARMONY(Hackathon on Resources for Modeling in Biology), as well as tutorials at the Inter-national Conference on Systems Biology (ICSB) and related meetings.
Contact us
Website: http://co.mbine.org/Get in touch with the editors: [email protected]
vi
Contents
Welcome to Heidelberg! iii
Funding is gratefully provided by: . . . . . . . . . . . . . . . . . . . . . . . iv
Welcome by the HITS management . . . . . . . . . . . . . . . . . . . . . . v
About HITS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Conference venue 1
Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Room names and location . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Conference Dinner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Meeting Agenda 3
Keynote Talks 13
Computational Physiology and the Physiome Project (Peter Hunter) . . 13
Implementing Open Science in publishing (Thomas Lemberger) . . . . . 13
Modeling projects across platforms - the reality and how reality should be(Ursula Kummer) . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Let’s go on a FAIR asset management safari (Carole Goble) . . . . . . . 14
From data via models to personalized medicine - An example in PediatricOncology (Norbert Graf ) . . . . . . . . . . . . . . . . . . . . . . . 15
Invited Talks 17
The History of COMBINE (Michael Hucka) . . . . . . . . . . . . . . . 17
How (not) to Apply (COMBINE) Standards in Agent-Based Modeling(Judith Wodke, Jens Hahn, Jorin Diemer, Edda Klipp) . . . . . . . 17
Computational modeling of liver function tests - Stratification and in-dividualization based on semantic annotations of models and data(Matthias Konig, Jan Grzegorzewski) . . . . . . . . . . . . . . . . 18
Research software engineers and their role for open and reproducible re-search (Heidi Seibold) . . . . . . . . . . . . . . . . . . . . . . . . . 19
A Standard Enabled Workflow for Synthetic Biology (Chris Myers) . . . 20
BIOROBOOST: A last chance for standardisation in Biology? (ManuelPorcar Miralles) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
The Center for Reproducible Biomedical Modeling (Herbert Sauro,MichaelL. Blinov, John H. Gennari, Arthur P. Goldberg, Jonathan R. Karr,Ion I.Moraru, David P. Nickerson) . . . . . . . . . . . . . . . . . . . 21
vii
A personalized modeling pipeline for cardiac electrophysiology simulationsof cardiac resynchronization therapy in infarct patients (CarolineMendonca Costa, Aurel Neic, Eric Kerfoot, Bradley Porter, Ben-jamin Sieniewicz, Justin Gould, Baldeep Sidhu, Zhong Chen, GernotPlank, Christopher A. Rinaldi, Martin J. Bishop, Steven A. Niederer) 22
CellDesigner: A modeling tool for biochemical networks (Akira Funa-hashi) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Virtual Cell: Modeling and Visualization of Reaction Rules (MichaelBlinov) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
EU-STANDS4PM: A pan-European Expert Forum joined forces to tacklethe complexity of big data integration for in silico methodologies inpersonalised medicine (Marc Kirschner) . . . . . . . . . . . . . . . 25
Standards in Biobanking (Petr Holub) . . . . . . . . . . . . . . . . . . . 25
Standards for Clinical Data and Systems Medicine (Søren Brunak) . . . 26
A Bird’s Eye view of Legal and Ethical aspects of EU-STANDS4PM (Katha-rina O Cathaoir) . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
In silico medicine and the VPH Institute: bringing the community togetherand the field forward (Liesbet Geris) . . . . . . . . . . . . . . . . . 27
Semantic annotation in the Physiome Model Repository (David Nicker-son, Dewan Sarwar, Tommy Yu, Anand Rampadarath, Koray Atalag) 28
Gene Ontology Causal Activity Modeling (GO-CAM) (Huaiyu Mi, PaulD. Thomas, David P. Hill, David Osumi-Sutherland, Kimberly VanAuken, Seth Carbon, James P. Balhoff, Laurent-Philippe Albou, Ben-jamin Good, Pascale Gaudet, Nomi L. Harris, Suzanna E. Lewis,Christopher J. Mungall) . . . . . . . . . . . . . . . . . . . . . . . . . 28
Sharing standardised models and data on the Open Source Brain platform(Padraig Gleeson) . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Multi-method modelling of synaptic plasticity and the challenges it brings(Melanie Stefan) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Contributed Talks 31
Analyzing Genetic Circuits for Hazards and Glitches (Pedro Fontanar-rosa, Hamid Hosseini; Amin Borujeni, Yuval Dorfan, Chris Voigt,Chris Myers) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
SBOL and its applicability in partially and fully automated design work-flows: Three success stories (Ernst Oberortner) . . . . . . . . . . . 31
Physiome - Publish your models curated for reproducibility and reusabilityto increase research quality for everyone (Karin Lundengard, PeterHunter, David Nickerson) . . . . . . . . . . . . . . . . . . . . . . . . 32
Reproducibility in model construction, validation and analysis workflowsin systems biology projects; Xylose metabolism in Caulobacter cres-centus as a case study, using JWS Online and the FAIRDOMHub(Jacky Snoep, Lu Shen, Bertina Siebers, Dawie van Niekerk, Wolf-gang Muller, Carole Goble ) . . . . . . . . . . . . . . . . . . . . . . . 33
viii
Causal, Mechanistic Pathway Based Analysis of -Omic Profiles (EmekDemir, Ozgun Babur, Funda Durupinar Babur, Hannah Manning,Michal Grzadkowski, Joseph Estabrook, Olga Nikolova, Kevin Watanabe-Smith) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Predictive in-silico multiscale analytics to support cancer personalized di-agnosis and prognosis, empowered by imaging biomarkers (PolyxeniGkontra, Leonor Cerd Alberich, Gracia Mart Besa, Luis Mart Bon-mat ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Datanator: Tools for Aggregating Data for Large-Scale Biomodeling (YosefRoth, Zhouyang Lian, Saahith Pochiraju, Jonathan Karr) . . . . . . 36
Simulation and Sensitivity Analysis for Large Kinetic Models in AMICI(Fabian Frohlich ) . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
OpenCOR: current status and future plans (Alan Garny, David Brooks,Peter Hunter) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Visualization of Part Use in SynBioHub (Jeanet Mante, Zach Zundel,Chris Myers) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
PySB framework: Tools to build, calibrate and visualize biochemical mod-els (Oscar O. Ortega, Carlos F. Lopez ) . . . . . . . . . . . . . . . 38
Visualization, Access, and Exploration of Biological Pathway Informationfrom Pathway Commons (Augustin Luna, Emek Demir, Igor Rod-chenkov, Ozgun Babur, Jeffrey Wong, Chris Sander, Gary Bader) . . 38
ELIXIR Cloud & AAI: Standardised and Interoperable Services for Hu-man Data Communities (Susheel Varma, Shubham Kapoor, AniaNiewielska, Alexander Kanitz, Jinny Chien, Andrea Cristofori, SarahButcher, Foivos Gypas, Kevin Sayers, Juha Tornroos, Marco Tan-garo, Giacinto Donvito, Yasset Perez-Riverol, Bjorn Gruning, Jen-nifer Harrow, Jonathan Tedds, Salvador Capella, Tommi Nyronen,Steven Newhouse) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
BioModels Parameters: A resource to search and access parameters frompublished systems models (Sheriff Rahuman, Chinmay Arankalle,Mihai Glont; Tung Nguyen, Henning Hermjakob) . . . . . . . . . . . 41
Two universes one world: Community standards vs. formal standards insystems biology and systems medicine (Martin Golebiewski ) . . . 41
Model curation and annotation (Anand Rampadarath) . . . . . . . . . 42
Identifiers.org Compact Identifiers for robust data citation (Henning Her-mjakob, Sarala Wimalaratne, Manuel Bernal Llinares, Javier Fer-rer, Nick Juty) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Principles for declarative multicellular modelling (Jorn Starruß, Walterde Back, Martin Golebiewski, Lutz Brusch ) . . . . . . . . . . . . . . 43
Flapjack: an open-source tool for storing, visualising, analysing and mod-elling kinetic gene expression data (Guillermo Yanez, Isaac Nunez,Tamara Matute, Fernan Federici, Timothy Rudge) . . . . . . . . . . . 43
Workshops and Breakouts 45
Sharing experiences in building a standards community (Dagmar Wal-temath, Mike Hucka ) . . . . . . . . . . . . . . . . . . . . . . . . . . 45
ix
SED-ML Script: a Proposal (Lucian Smith, Herbert Sauro) . . . . . . . . . 45
COMBINE as Legal Entity (Martin Golebiewski, Herbert Sauro ) . . . . . 46
SBGN workshop (Falk Schreiber, Michael Blinov ) . . . . . . . . . . . . . 47
MIRIAM 2 & the OMEX Metadata Specification (Dagmar Waltemath,David Nickerson) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
SBOL breakout (Ernst Oberortner, Chris Myers ) . . . . . . . . . . . . . . 48
Model eXchange consortium: Inaugural meeting (Rahuman Sheriff, Hen-ning Hermjakob) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
EU-STANDS4PM workshop (Marc Kirschner, Martin Golebiewski ) . . . . 49
Something old, something new: constraint-based modelling and SBML(Brett G. Olivier, Frank T. Bergmann) . . . . . . . . . . . . . . . . . 50
SBML L3 Packages - An Introduction to SBML packages (Sarah Keating) 50
FAIRDOM PALs and User Workshop (Olga Krebs ) . . . . . . . . . . . . . 51
Model eXchange consortium: Inaugural meeting (Rahuman Sheriff, Hen-ning Hermjakob) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Lightning Talks 53
Systems Biology Graphical Notation (SBGN) (Michael Blinov) . . . . . 53
SBML Level 3 Version 2 and SBML Level 3 packages (Michael Hucka) . 53
WebProv: A web-based tool to access, store, and display provenance in-formation of simulation models (Kai Budde, Jacob Smith, AndreasRuscheinski, Adelinde M. Uhrmacher) . . . . . . . . . . . . . . . . . 54
MIRIAM 2 & the OMEX Metadata Specification (Dagmar Waltemath,David Nickerson) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
A Virtual Cohort of Heart Failure Patients Four-chamber Heart Meshes forCardiac Electro-mechanics Simulations (Marina Strocchi, ChristophAugustin, Matthias Gsell, Orod Razeghi, Anton Prassl, Edward Vig-mond, Jonathan Behar, Justin Gould, Sidhu Baldeep, Aldo Rinaldi,Martin Bishop, Gernot Plank, Steven Niederer) . . . . . . . . . . . . 55
WebLab update (Sarah Keating) . . . . . . . . . . . . . . . . . . . . . . 55
NESTML: A domain-specific language for biologically realistic neuron andsynapse models (Charl Linssen , Jochen M. Eppler, Abigail Mor-rison) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
BioModels 2019 (Mihai Glont, Henning Hermjakob, Rahuman Sheriff,Krishna Tiwari, Tung Nguyen, Manh Tu Vu) . . . . . . . . . . . . . 57
Minibatch optimization: a method for training mechanistic models usinglarge data sets (Paul Stapor, Leonard Schmiester, Jan Hasenauer,Daniel Weindl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Modelling pancreas - Computational modelling and analysis of alteredmetabolism in the pancreas and its relationship to disease (DeepaMaheshvare Soumyendu Raha, Matthias Konig, Debnath Pal ) . . . 58
PK-DB 1.0: a pharmacokinetics database for individualized and stratifiedmodeling (Jan Grzegorzewski, Matthias Konig) . . . . . . . . . . 58
Standardised meta-data in Hospital Information Systems unlock clinicaldata for research purposes A real-world example (Robert Gott, KaiFitzer, Torsten Leddig, Dagmar Waltemath, Thomas Bahls) . . . 59
x
Development of python visualization package DNAplotlib (Sunwoo Kang) 60
Cy3SBML: a Cytoscape app for SBML (Matthias Konig, Nicolas Ro-driguez, Andreas Drager) . . . . . . . . . . . . . . . . . . . . . . . . . 60
Posters 61
Review of software tools and data sources supporting the Systems Biol-ogy Graphical Notation (SBGN) standard (Michael Blinov, AdrienRougny, Andreas Drager, Vasundra Toure, Ugur Dogrusoz, Alexan-der Mazein) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
WebProv: A web-based tool to access, store, and display provenance in-formation of simulation models (Kai Budde, Jacob Smith, AndreasRuscheinski, Adelinde M. Uhrmacher) . . . . . . . . . . . . . . . . . 61
Comprehensive mathematical modeling of sphingolipid metabolism by in-tegration of lipidomics and proteomics data (Canan Has, ThorstenHornemann, Robert Ahrends ) . . . . . . . . . . . . . . . . . . . . . . 62
The Systems Biology Graphical Notation: a standardised representationof biological maps (Andreas Drager, Vasundra Toure, AlexanderMazein, Adrien Rougny, Ugur Dogrusoz, Michael Blinov, AugustinLuna) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
AMICI and PESTO - A strong couple for simulation and parameter es-timation in computational and systems biology (Erika Dudkin,Paul Stapor, Daniel Weindl, Benjamin Ballnus, Sabine Hug, CarolinLoos, Anna Fiedler, Sabrina Krause, Sabrina Hroß, Yannik Schalte,Leonard Schmiester, Dantong Wang, Elba Raimndez Alvarez, SimonMerkt, Fabian Frohlich, Jan Hasenauer) . . . . . . . . . . . . . . . . 63
Analyzing Genetic Circuits for Hazards and Glitches (Pedro Fontanar-rosa, Hamid Hosseini; Amin Borujeni, Yuval Dorfan, Chris Voigt,Chris Myers) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
YoMoSt - Statistics for you local model versions (Tom Gebhardt, MartinScharm, Olaf Wolkenhauer) . . . . . . . . . . . . . . . . . . . . . . . 65
PK-DB 1.0: a pharmacokinetics database for individualized and stratifiedmodeling (Jan Grzegorzewski, Matthias Konig) . . . . . . . . . . 65
EnzymeML - a SBML-based exchange format for biocatalytic data (ColinHalupczok, Jurgen Pleiss) . . . . . . . . . . . . . . . . . . . . . . . 65
The JSBML project: a fully featured Java API for working with sys-tems biology models (Thomas M. Hamm, Nicolas Rodriguez, Ro-man Schulte, Leandro Watanabe, Ibrahim Y. Vazirabad, Victor Kofia,Chris J. Myers, Akira Funahashi, Michael Hucka, Andreas Drager ) . 66
Identifiers.org Compact Identifiers for robust data citation (Henning Her-mjakob, Sarala Wimalaratne, Manuel Bernal Llinares, Javier Fer-rer, Nick Juty) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Experiences with Developing Digital Registries for Contraceptive Care us-ing the Collaborative Requirements Development Methodology (Es-ther Inau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Virtual patient (Fedor Kolpakov) . . . . . . . . . . . . . . . . . . . . . . 68
xi
FAIRDOM approach for Systems Biology Data and Model Management(Olga Krebs, Martin Golebiewski, Stuart Owen, Maja Rey, NatalieStanford, Ulrike Wittig, Xiaoming Hu, Katy Wolstencroft, Jacky L.Snoep, Bernd Rinn, Wolfgang Muller, Carole Goble) . . . . . . . . . 69
NESTML: A domain-specific language for biologically realistic neuron andsynapse models (Charl Linssen , Jochen M. Eppler, Abigail Mor-rison) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Physiome - Publish your models curated for reproducibility and reusabilityto increase research quality for everyone (Karin Lundengard, PeterHunter, David Nickerson) . . . . . . . . . . . . . . . . . . . . . . . . 70
Modelling pancreas - Computational modelling and analysis of alteredmetabolism in the pancreas and its relationship to disease (DeepaMaheshvare Soumyendu Raha, Matthias Konig, Debnath Pal ) . . . 70
Enhancing the computational based function annotation of thiamine diphos-phate dependent enzymes (Stephan Malzacher, Patrick Buchholz,Colin Halupczok, Saskia Bock, Dorte Rother, Jurgen Pleiss) . . . . . 70
Visualization of Part Use in SynBioHub (Jeanet Mante, Zach Zundel,Chris Myers) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Comprehensive Modelling Platform (Lukrecia Mertova, Matej Trojak,Jan Cerveny, Marek Havlı, Matej Hajna, Radoslav Doktor, OndrejLostak) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
PySB framework: Tools to build, calibrate and visualize biochemical mod-els (Oscar O. Ortega, Carlos F. Lopez ) . . . . . . . . . . . . . . . 72
The Center for Reproducible Biomedical Modeling (Anand Rampadarath,David Nickerson, Michael Blinov, John Gennari, Arthur Goldberg,Jonathan Karr, Herbert Sauro) . . . . . . . . . . . . . . . . . . . . . 72
Datanator: Tools for Aggregating Data for Large-Scale Biomodeling (YosefRoth, Zhouyang Lian, Saahith Pochiraju, Jonathan Karr) . . . . . . 72
Standardizing Electronic Laboratory Notebooks to improve their integra-tion with data from the biomedical domain (Max Schroder, Dag-mar Waltemath, Frank Kruger, Sascha Spors) . . . . . . . . . . . . . 73
Minibatch optimization: a method for training mechanistic models usinglarge data sets (Paul Stapor, Leonard Schmiester, Jan Hasenauer,Daniel Weindl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
A Virtual Cohort of Heart Failure Patients Four-chamber Heart Meshes forCardiac Electro-mechanics Simulations (Marina Strocchi, ChristophAugustin, Matthias Gsell, Orod Razeghi, Anton Prassl, Edward Vig-mond, Jonathan Behar, Justin Gould, Sidhu Baldeep, Aldo Rinaldi,Martin Bishop, Gernot Plank, Steven Niederer) . . . . . . . . . . . . 74
SABIO-RK: kinetic data for systems biology (Andreas Weidemann,Ulrike Wittig, Maja Rey, Wolfgang Muller) . . . . . . . . . . . . . . . 74
Flapjack: an open-source tool for storing, visualising, analysing and mod-elling kinetic gene expression data (Guillermo Yanez, Isaac Nunez,Tamara Matute, Fernan Federici, Timothy Rudge) . . . . . . . . . . . 74
xii
Standardised meta-data in Hospital Information Systems unlock clinicaldata for research purposes A real-world example (Robert Gott, KaiFitzer, Torsten Leddig, Dagmar Waltemath, Thomas Bahls) . . . 75
xiii
xiv
Conference venue
COMBINE 2019 will be held at the conference facilities of Studio Villa Bosch in Hei-delberg. The address is Schloss-Wolfsbrunnenweg 33, 69118 Heidelberg, Germany.
Directions
• By plane: You travel to the international airport of Frankfurt/Main. Fromthere, you take the Lufthansa-Bus (leaving in front of terminal 1, section B) toHeidelberg or Mannheim/Heidelberg (depending on the time of the day). Inapproximately one hour it will bring you to the Crowne Plaza in Heidelberg.From there you can take a taxi for a 15 minute ride to Studio Villa Bosch inthe Schloss-Wolfsbrunnenweg.
• By train: You travel to Heidelberg Hauptbahnhof. From there you can eithertake public transport, as mentioned below, or a taxi to the Studio Villa Bosch.
• By bus: You can reach HITS and Studio Villa Bosch directly with the busnumber 30, the so called science-bus, which runs hourly from Monday to Fri-day. At Heidelberg main station you take the train S2 (direction Eberbach),S1 (direction Osterburken) or S5 (direction Eppingen Bahnhof) and get off atHeidelberg-Altstadt station. There you change to bus number 30, directionSchlierbach (HD), HITS. Please get off the bus at the bus stop ”Villa Bosch”.
At the main station you can also take:
– bus number 32 (direction Universittsplatz). You get off at the bus stopUniversittsplatz and change to bus number 30, direction Schlierbach(HD), HITS.
– bus number 33 (direction Ziegelhausen, Kpfel). You leave at the bus stopPeterskirche and change to bus number 30, direction Schlierbach (HD),HITS.
• By car: You travel on the Autobahn to the Heidelberger Kreuz (Heidelbergcrossing), from there on route A 656 directly to Heidelberg. At the end ofthe Autobahn you turn left and then right; you are now on B 37, whichyou will follow some minutes, until Heidelberg downtown ends (ca. 5 km).After a few hundred meters, you will see a pointer to HITS, which leads youupwards to the right, across the railway. Then turn left and follow the streetsAm Rosenbusch and Hausackerweg, always climbing upwards the serpentine
1
curves. The Hausackerweg ends by joining the Schloss-Wolfsbrunnenweg atthe Hotel Atlantic. Turn left, following the HITS pointer onto the Schloss-Wolfsbrunnenweg. 300 m further at your left you have reached the gatewayto the Villa Bosch. The HITS campus is about 200 meters farther, on theleft-hand side. Parking: in the garage Unter der Boschwiese, opposite to theVilla Bosch.
Room names and location
Conference Dinner
Restaurant ’S’ Kastanie, Elisabethenweg 1 in HeidelbergWednesday, July 17th, 2019 - 19:00 - 23:00Registered accompanying persons (payable IN CASH at the restaurant): 70 Euro
2
Meeting Agenda
3
Mon
day,
July
15t
h 20
19
time
Plen
ary
Room
: Fo
yer
09:1
5-10
:00
Regi
stra
tion
& C
offe
e10
:00
– 11
:00
10:0
0 - 1
0:05
Mar
tin G
oleb
iew
ski, H
ITS,
Hei
delb
erg
(D)
Wel
com
e by
the
host
10:0
5 - 1
0:20
Gesa
Sch
önbe
rger
and
Wol
fgan
g M
ülle
r, HI
TS,
Heid
elbe
rg (D
)W
elco
me
addr
esse
s fro
m th
e HI
TS M
anag
emen
t
10:2
0 - 1
1:00
Mich
ael H
ucka
, Cal
iforn
ia In
stitu
te o
f Te
chno
logy
, Pas
aden
a, C
A (U
SA)
The
Hist
ory
of C
OM
BINE
(inv
ited
talk
)
11:0
0 –
12:0
011
:00
- 12:
00Pe
ter H
unte
r (Au
ckla
nd, N
ew Z
eala
nd)
Keyn
ote:
Com
puta
tiona
l Phy
siolo
gy a
nd th
e Ph
ysio
me
Proj
ect
12:0
0 - 1
3:00
Lunc
h
13:0
0 - 1
3:45
Thom
as Le
mbe
rger
(EM
BO p
ress
, D)
Keyn
ote:
Impl
emen
ting
Ope
n Sc
ienc
e in
pub
lishi
ng
13:4
5 - 1
4:30
Urs
ula
Kum
mer
, Uni
vers
ity o
f Hei
delb
erg
(D)
Keyn
ote:
Mod
elin
g pr
ojec
ts a
cros
s pla
tform
s - th
e re
ality
and
how
real
ity
shou
ld b
e
14:3
0 - 1
5:00
Light
ning
talk
s (1)
3 m
in ta
lks s
elec
ted
from
the
subm
itted
abs
tract
s
15:0
0 - 1
5:30
Post
er S
essio
n &
Cof
fee
Brea
k
15:3
0 - 1
6:00
Judi
th W
odke
, Hum
bold
t Uni
vers
ity B
erlin
(D)
How
(not
) to
Appl
y (C
OM
BINE
) Sta
ndar
ds in
Age
nt-B
ased
Mod
elin
g16
:00
- 16:
30M
atth
ias K
önig
, Hum
bold
t Uni
vers
ity B
erlin
(D)
Com
puta
tiona
l mod
elin
g of
liver
func
tion
test
s - S
tratif
icatio
n an
d in
divi
dual
izatio
n ba
sed
on se
man
tic a
nnot
atio
ns o
f mod
els a
nd d
ata
16:3
0 - 1
7:00
Heid
i Sei
bold
, Lud
wig
-Max
imilia
ns-U
nive
rsitä
t M
ünch
en (D
)Re
sear
ch so
ftwar
e en
gine
ers a
nd th
eir r
ole
for o
pen
and
repr
oduc
ible
re
sear
ch
17:0
0 - 1
7:30
Light
ning
talk
s (2)
3 m
in ta
lks s
elec
ted
from
the
subm
itted
abs
tract
s
17:3
0 - 1
9:00
de.N
BI C
ollo
quiu
m; C
hair:
Dag
mar
Wal
tem
ath
(Gre
ifsw
ald,
D)
Post
er S
essio
n &
Wel
com
e Re
cept
ion:
10
year
s CO
MBI
NE
- Hap
py B
irthd
ay !
Carl
Bosc
h Au
dito
rium
(H1)
Ope
ning
Ses
sion;
Cha
ir: M
artin
Gol
ebie
wsk
i (He
idel
berg
, D)
HITS
Col
loqu
ium
; Cha
ir: W
olfg
ang
Mül
ler (
Heid
elbe
rg, D
)
COM
BIN
E Co
lloqu
ium
; Cha
ir: M
artin
Gol
ebie
wsk
i (He
idel
berg
, D)
13:0
0 –
15:0
0
15:3
0 –
17:3
0
Tues
day,
July
16t
h 20
19
time
Plen
ary/
Brea
kout
sess
ion
1Br
eako
ut se
ssio
n 2
Brea
kout
sess
ion
3Ro
om:
Foye
rFr
itz H
aber
Aud
itoriu
m (H
2)Al
win
Mitt
asch
Stu
dio
(H3)
09:3
0 –
11:1
0Se
ssio
n 1:
09
:30
- 10:
00Ch
ris M
yers
, Uni
vers
ity o
f Uta
h,
Salt
Lake
City
, UT
(USA
)A
Stan
dard
Ena
bled
Wor
kflo
w fo
r Syn
thet
ic Bi
olog
y (in
vite
d ta
lk)
10:0
0 - 1
0:30
Man
uel P
orca
r Mira
lles,
Inst
itute
fo
r Int
egra
tive
Syst
ems B
iolo
gy,
Valè
ncia
(Spa
in)
BioR
oboo
st: A
last
chan
ce fo
r sta
ndar
disa
tion
in
biol
ogy?
(inv
ited
talk
)
10:3
0 - 1
0:40
Pedr
o Fo
ntan
arro
sa, U
nive
rsity
of
Utah
, Sal
t Lak
e Ci
ty, U
T (U
SA)
Anal
yzin
g Ge
netic
Circ
uits
for H
azar
ds a
nd G
litch
es
10:4
0 - 1
0:50
Erns
t Obe
rortn
er, L
awre
nce
Berk
eley
Nat
. Lab
orat
ory
(USA
)SB
OL a
nd it
s app
licab
ility
in p
artia
lly a
nd fu
lly
auto
mat
ed d
esig
n w
orkf
low
s: T
hree
succ
ess s
torie
s
10:5
0 - 1
1:05
all s
peak
ers o
f Ses
sion
1Di
scus
sion
11:0
5 - 1
1:30
Post
er S
essio
n &
Cof
fee
11:3
0 –
13:0
0Se
ssio
n 2:
11
:30
- 12:
00He
rber
t Sau
ro, U
nive
rsity
of
Was
hing
ton,
Sea
ttle,
WA
(USA
)Th
e Ce
nter
for R
epro
ducib
le B
iom
edica
l Mod
elin
g
(invi
ted
talk
)12
:00
- 12:
10Ka
rin Lu
nden
gård
, Uni
vers
ity o
f Au
ckla
nd (N
Z)Ph
ysio
me
- Pub
lish
your
mod
els c
urat
ed fo
r re
prod
ucib
ility
and
reus
abilit
y to
incr
ease
rese
arch
qu
ality
for e
very
one
12:1
0 - 1
2:20
Jack
y Sn
oep,
Ste
llenb
osch
Un
iver
sity
(Sou
th A
frica
)Re
prod
ucib
ility
in m
odel
cons
truct
ion,
val
idat
ion
and
anal
ysis
wor
kflo
ws i
n sy
stem
s bio
logy
pro
ject
s;
Xylo
se m
etab
olism
in C
aulo
bact
er cr
esce
ntus
as a
ca
se st
udy,
usin
g JW
S O
nlin
e an
d th
e 12
:20
- 12:
30Em
ek D
emir,
Ore
gon
Heal
th a
nd
Scie
nce
Univ
ersit
y, P
ortla
nd (U
SA)
Caus
al, M
echa
nist
ic Pa
thw
ay B
ased
Ana
lysis
of -
Om
ic Pr
ofile
s12
:30
- 12:
40Po
lyxe
ni G
kont
ra, L
a Fe
Hea
lth
Rese
arch
Inst
itute
(IIS
La F
e),
Vale
ncia
(Spa
in)
Pred
ictiv
e in
-silic
o m
ultis
cale
ana
lytic
s to
supp
ort
canc
er p
erso
naliz
ed d
iagn
osis
and
prog
nosis
, em
pow
ered
by
imag
ing
biom
arke
rs12
:40
- 13:
00al
l spe
aker
s of S
essio
n 2
Disc
ussio
n
Repr
oduc
ibilit
y in
Syn
thet
ic an
d Sy
stem
s Bio
logy
; Cha
irs: D
agm
ar W
alte
mat
h (G
reifs
wal
d, D
) & D
avid
Nick
erso
n (A
uckl
and,
NZ)
Carl
Bosc
h Au
dito
rium
(H1)
Synt
hetic
Bio
logy
; Cha
irs: E
rnst
Obe
rortn
er (B
erke
ley,
CA,
USA
) & C
hris
Mye
rs (S
alt L
ake
City
, UT,
USA
)
time
Plen
ary/
Brea
kout
sess
ion
1Br
eako
ut se
ssio
n 2
Brea
kout
sess
ion
3Ro
om:
Foye
rFr
itz H
aber
Aud
itoriu
m (H
2)Al
win
Mitt
asch
Stu
dio
(H3)
13:0
0 - 1
4:00
Lunc
h14
:00
– 15
:30
Sess
ion
3:
Repr
oduc
ibilit
y - S
ED-M
L Sc
ript:
a Pr
opos
al; C
hairs
: He
rber
t Sau
ro &
Lucia
n Sm
ith (S
eattl
e, W
A, U
SA)
15:3
0 - 1
6:00
Post
er S
essio
n &
Cof
fee
Brea
k16
:00
– 17
:30
Sess
ion
4:
17:3
0 - 1
7:45
Plen
ary
Wra
pup
18:3
0 - 2
0:30
COM
BIN
E as
Lega
l Ent
ity; C
hairs
: Mar
tin G
oleb
iew
ski (
Heid
elbe
rg, D
) & H
erbe
rt Sa
uro
(Sea
ttle,
WA,
USA
)
Carl
Bosc
h Au
dito
rium
(H1)
Guid
ed T
our t
hrou
gh th
e ol
d to
wn
of H
eide
lber
g (o
ptio
nal)
Shar
ing
expe
rienc
es in
bui
ldin
g a
stan
dard
s com
mun
ity; C
hairs
: Dag
mar
Wal
tem
ath
(Gre
ifsw
ald,
D) &
Mik
e Hu
cka
(Pas
aden
a, C
A, U
SA)
Wed
nesd
ay, J
uly
17th
201
9
time
Plen
ary/
Brea
kout
sess
ion
1Br
eako
ut se
ssio
n 2
Brea
kout
sess
ion
3Ro
om:
Foye
rFr
itz H
aber
Aud
itoriu
m (H
2)Al
win
Mitt
asch
Stu
dio
(H3)
09:3
0 –
11:0
0Se
ssio
n 5:
09
:30
- 10:
00Ca
rolin
e M
endo
nça
Cost
a, K
ing'
s Co
llege
Lond
on (U
K)A
pers
onal
ized
mod
elin
g pi
pelin
e fo
r car
diac
el
ectro
phys
iolo
gy si
mul
atio
ns o
f car
diac
re
sync
hron
izatio
n th
erap
y in
infa
rct p
atie
nts
(invi
ted
talk
)
10:0
0 - 1
0:10
Yose
f Rot
h, Ic
ahn
Scho
ol o
f M
edici
ne a
t Mou
nt S
inai
, New
Yo
rk C
ity, N
Y (U
SA)
Data
nato
r: To
ols f
or A
ggre
gatin
g Da
ta fo
r Lar
ge-
Scal
e Bi
omod
elin
g
10:1
0 - 1
0:20
Fabi
an F
röhl
ich, H
arva
rd M
edica
l Sc
hool
, Bos
ton,
MA
(USA
)Si
mul
atio
n an
d Se
nsiti
vity
Ana
lysis
for L
arge
Kin
etic
Mod
els i
n AM
ICI
10:2
0 - 1
0:30
Alan
Gar
ny, U
nive
rsity
of A
uckl
and
(NZ)
Ope
nCO
R: cu
rrent
stat
us a
nd fu
ture
pla
ns
10:3
0 - 1
0:45
all s
peak
ers o
f Ses
sion
5Di
scus
sion
10:4
5 - 1
1:15
Post
er S
essio
n &
Cof
fee
11:1
5 –
13:0
0Se
ssio
n 6:
11
:15
- 11:
45Ak
ira F
unah
ashi
, Kei
o Un
iver
sity,
Yo
koha
ma
(Japa
n)Ce
llDes
igne
r: A
mod
elin
g to
ol fo
r bio
chem
ical
netw
orks
(inv
ited
talk
)
11:4
5 - 1
2:15
Mich
ael B
linov
, UCo
nn H
ealth
, Fa
rmin
gton
, CT
(USA
)Vi
rtual
Cel
l: M
odel
ing
and
Visu
aliza
tion
of R
eact
ion
Rule
s (in
vite
d ta
lk)
12:1
5 - 1
2:25
Jean
et M
ante
, Uni
vers
ity o
f Uta
h,
Salt
Lake
City
, UT
(USA
)Vi
sual
izatio
n of
Par
t Use
in S
ynBi
oHub
12:2
5 - 1
2:35
Osc
ar O
Orte
ga, V
ande
rbilt
Un
iver
sity,
Nas
hville
, TN
(USA
)Py
SB fr
amew
ork:
Too
ls to
bui
ld, c
alib
rate
and
vi
sual
ize b
ioch
emica
l mod
els
12:3
5 - 1
2:45
Augu
stin
Luna
, Har
vard
Med
ical
Scho
ol, B
osto
n, M
A (U
SA)
Visu
aliza
tion,
Acc
ess,
and
Expl
orat
ion
of B
iolo
gica
l Pa
thw
ay In
form
atio
n fro
m P
athw
ay C
omm
ons
12:4
5 - 1
3:00
all s
peak
ers o
f Ses
sion
6Di
scus
sion
13:0
0 - 1
4:00
Lunc
h
Carl
Bosc
h Au
dito
rium
(H1)
Visu
aliza
tion;
Cha
irs: F
alk
Schr
eibe
r (Ko
nsta
nz, D
) & M
ichae
l Blin
ov (F
arm
ingt
on, C
T, U
SA)
Mod
elin
g Ap
proa
ches
and
Too
ls; C
hairs
: Mat
thia
s Kön
ig (B
erlin
, D) &
Mel
anie
Ste
fan
(Edi
nbur
gh, U
K)
time
Plen
ary/
Brea
kout
sess
ion
1Br
eako
ut se
ssio
n 2
Brea
kout
sess
ion
3Ro
om:
Foye
rFr
itz H
aber
Aud
itoriu
m (H
2)Al
win
Mitt
asch
Stu
dio
(H3)
14:0
0 –
15:3
0Se
ssio
n 7:
SB
GN w
orks
hop;
Cha
irs: F
alk
Schr
eibe
r (Ko
nsta
nz, D
) & M
ichae
l Bl
inov
(Far
min
gton
, CT,
USA
)
EV
ERYB
ODY
IS IN
VITE
D TO
AT
TEND
http
s://
sbgn
.gith
ub.io
/sbg
n12
MIR
IAM
2 &
the
OM
EX
Met
adat
a Sp
ecifi
catio
n;
Chai
rs: D
agm
ar W
alte
mat
h (G
reifs
wal
d, D
) & D
avid
Ni
cker
son
(Auc
klan
d, N
Z)
SBO
L; C
hairs
: Ern
st
Obe
rortn
er (B
erke
ley,
CA,
US
A) &
Chr
is M
yers
(Sal
t La
ke C
ity, U
T, U
SA)
15:3
0 - 1
6:00
Post
er S
essio
n &
Cof
fee
Brea
k16
:00
– 17
:30
Sess
ion
8:SB
GN w
orks
hop;
Cha
irs: F
alk
Schr
eibe
r (Ko
nsta
nz, D
) & M
ichae
l Bl
inov
(Far
min
gton
, CT,
USA
)
EV
ERYB
ODY
IS IN
VITE
D TO
AT
TEND
http
s://
sbgn
.gith
ub.io
/sbg
n12
Mod
el e
Xcha
nge
cons
ortiu
m: I
naug
ural
m
eetin
g; C
hairs
: Hen
ning
He
rmja
kob
& R
ahum
an
Sher
iff (E
MBL
-EBI
, Hin
xton
, UK
)
SBO
L; C
hairs
: Ern
st
Obe
rortn
er (B
erke
ley,
CA,
US
A) &
Chr
is M
yers
(Sal
t La
ke C
ity, U
T, U
SA)
17:3
0 - 1
7:45
Plen
ary
Wra
pup
19:3
0 - 2
3:00
Carl
Bosc
h Au
dito
rium
(H1)
Conf
eren
ce D
inne
r at t
he re
stau
rant
S'K
asta
nie
(add
ress
: Elis
abet
henw
eg 1
in H
eide
lber
g): h
ttps:
//w
ww
.res
taur
ant-s
-kas
tani
e.de
Thur
sday
, Jul
y 18
th 2
019
time
Plen
ary/
Brea
kout
sess
ion
1Br
eako
ut se
ssio
n 2
Brea
kout
sess
ion
3Ro
om:
Foye
rFr
itz H
aber
Aud
itoriu
m (H
2)Al
win
Mitt
asch
Stu
dio
(H3)
09:3
0 –
11:0
0Se
ssio
n 9:
09
:30
- 10:
15Ca
role
Gob
le, U
nive
rsity
of
Man
ches
ter (
UK)
Keyn
ote:
Let’s
go
on a
FAI
R as
set m
anag
emen
t sa
fari
10:1
5 - 1
0:25
Sush
eel V
arm
a, E
MBL
-EBI
, Hin
xton
(U
K)EL
IXIR
Clo
ud &
AAI
: Sta
ndar
dise
d an
d In
tero
pera
ble
Serv
ices f
or H
uman
Dat
a Co
mm
uniti
es
10:2
5 - 1
0:35
Rahu
man
She
riff,
EMBL
-EBI
, Hi
nxto
n (U
K)Bi
oMod
els P
aram
eter
s: A
reso
urce
to se
arch
and
ac
cess
par
amet
ers f
rom
pub
lishe
d sy
stem
s mod
els
10:3
5 - 1
0:45
Mar
tin G
oleb
iew
ski, H
ITS,
He
idel
berg
(D)
Two
univ
erse
s – o
ne w
orld
: Com
mun
ity st
anda
rds
vs. f
orm
al st
anda
rds i
n sy
stem
s bio
logy
and
sy
stem
s med
icine
10:4
5 - 1
1:00
all s
peak
ers o
f Ses
sion
9Di
scus
sion
11:0
0 - 1
1:30
Post
er S
essio
n &
Cof
fee
Brea
k11
:30
– 13
:00
Sess
ion
10:
11:3
0 - 1
1:45
Mar
c Kirs
chne
r, PT
J (Ge
rman
y)
and
Ingr
id S
kelto
n Ko
ckum
, Ka
rolin
ska
Inst
itute
t (Sw
eden
)
EU-S
TAND
S4PM
: A p
an-E
urop
ean
Expe
rt Fo
rum
jo
ined
forc
es to
tack
le th
e co
mpl
exity
of b
ig d
ata
inte
grat
ion
for i
n sil
ico m
etho
dolo
gies
in
pers
onal
ised
med
icine
11:4
5 - 1
2:15
Nor
bert
Gra
f, Sa
arla
nd U
nive
rsity
M
edica
l Cen
ter,
Hom
burg
(D)
Keyn
ote:
Fro
m d
ata
via
mod
els t
o pe
rson
alize
d m
edici
ne –
An
exam
ple
in P
edia
tric O
ncol
ogy
12:1
5 - 1
2:30
Petr
Hol
ub, B
BMRI
-ERI
C, M
asar
yk
Univ
ersit
y (C
zech
Rep
ublic
)St
anda
rds i
n Bi
oban
king
(inv
ited
talk
)
12:3
0 - 1
2:45
Søre
n Br
unak
, Uni
vers
ity o
f Co
penh
agen
(DK)
Stan
dard
s for
Clin
ical D
ata
and
Syst
ems M
edici
ne
12:4
5 - 1
3:00
Kath
arin
a Ev
a Ó
Cat
haoi
r, Un
iver
sity
of C
open
hage
n (D
K)A
Bird
's Ey
e vi
ew o
f Leg
al a
nd E
thica
l asp
ects
of
EU
-STA
NDS4
PM
Carl
Bosc
h Au
dito
rium
(H1)
Mod
el a
nd D
ata
Man
agem
ent;
Chai
rs: W
olfg
ang
Mül
ler &
Ulri
ke W
ittig
(Hei
delb
erg,
D)
Stan
dard
s for
Per
sona
lized
Med
icine
; Cha
irs: M
arc K
irsch
ner (
Jülic
h, D
) & M
artin
Gol
ebie
wsk
i (He
idel
berg
, D)
time
Plen
ary/
Brea
kout
sess
ion
1Br
eako
ut se
ssio
n 2
Brea
kout
sess
ion
3Ro
om:
Foye
rFr
itz H
aber
Aud
itoriu
m (H
2)Al
win
Mitt
asch
Stu
dio
(H3)
13:0
0 - 1
4:00
Lunc
hCO
MBI
NE
coor
dina
tors
' mee
ting
14:0
0 - 1
4:30
Lie
sbet
Ger
is, E
xecu
tive
Dire
ctor
VP
H In
stitu
te, K
U Le
uven
(Bel
gium
)In
silic
o m
edici
ne a
nd th
e VP
H in
stitu
te: b
ringi
ng th
e co
mm
unity
toge
ther
and
the
field
forw
ard
(invi
ted
talk
)
14:3
0 –
15:3
0Se
ssio
n 11
: EU
-STA
NDS
4PM
wor
ksho
p;
Chai
rs: M
arc K
irsch
ner (
Jülic
h, D
) &
Mar
tin G
oleb
iew
ski (
Heid
elbe
rg, D
)
EV
ERYB
ODY
IS IN
VITE
D TO
AT
TEND
Link
to th
e ag
enda
: ht
tp://
co.m
bine
.org
/sys
tem
/file
s/EU
-ST
AND
S4PM
_CO
MBI
NE_
wor
ksho
p_pu
blis
hed_
0.pd
f
15:3
0 - 1
6:00
Post
er S
essio
n &
Cof
fee
Brea
k16
:00
– 17
:30
Sess
ion
12:
EU-S
TAN
DS4P
M w
orks
hop;
Ch
airs
: Mar
c Kirs
chne
r (Jü
lich,
D) &
M
artin
Gol
ebie
wsk
i (He
idel
berg
, D)
EVER
YBO
DY IS
INVI
TED
TO
ATTE
ND
Link
to th
e ag
enda
: ht
tp://
co.m
bine
.org
/sys
tem
/file
s/EU
-ST
AND
S4PM
_CO
MBI
NE_
wor
ksho
p_pu
blis
hed_
0.pd
f
SBM
L L3
Pack
ages
- An
In
trod
uctio
n to
SBM
L pac
kage
s;
Chai
r: Sa
rah
Keat
ing
(Uni
vers
ity
Colle
ge Lo
ndon
, UK)
SBO
L; C
hairs
: Chr
is M
yers
&
Ern
st O
bero
rtner
17:3
0 - 1
7:45
Plen
ary
Wra
pup
SBO
L; C
hairs
: Chr
is M
yers
&
Ern
st O
bero
rtner
Som
ethi
ng o
ld, s
omet
hing
new
: co
nstr
aint
-bas
ed m
odel
ling
and
SBM
L; C
hair:
Bre
tt O
livie
r (VU
Un
iver
sity
Amst
erda
m, N
L)
Carl
Bosc
h Au
dito
rium
(H1)
Frid
ay, J
uly
19th
201
9
time
Plen
ary/
Brea
kout
sess
ion
1Br
eako
ut se
ssio
n 2
Brea
kout
sess
ion
3Ro
om:
Foye
rFr
itz H
aber
Aud
itoriu
m (H
2)Al
win
Mitt
asch
Stu
dio
(H3)
09:3
0 –
11:0
0Se
ssio
n 13
: 09
:30
- 10:
00Da
vid
Nick
erso
n, A
uckl
and
Bioe
ngin
eerin
g In
stitu
te, U
nive
rsity
of
Auc
klan
d (N
ew Z
eala
nd)
Sem
antic
ann
otat
ion
in th
e Ph
ysio
me
Mod
el
Repo
sitor
y (in
vite
d ta
lk)
10:0
0 - 1
0:30
Huai
yu M
i, Kec
k Sc
hool
of
Med
icine
of t
he U
nive
rsity
of
Sout
hern
Cal
iforn
ia (U
SC),
Los
Ange
les,
CA (U
SA)
Gene
Ont
olog
y Ca
usal
Act
ivity
Mod
elin
g (G
O-C
AM)
- inv
ited
talk
10:3
0 - 1
0:40
Anan
d Ra
mpa
dara
th, U
nive
rsity
of
Auck
land
(NZ)
Mod
el cu
ratio
n an
d an
nota
tion
10:4
0 - 1
0:50
Henn
ing
Herm
jako
b, E
MBL
-EBI
, Hi
nxto
n (U
K)Id
entif
iers
.org
Com
pact
Iden
tifie
rs fo
r rob
ust d
ata
citat
ion
10:5
0 - 1
1:00
all s
peak
ers o
f Ses
sion
13Di
scus
sion
11:0
0 –
11:3
0Po
ster
Ses
sion
& C
offe
e 11
:30
– 13
:00
Sess
ion
14:
11:3
0 - 1
1:50
Padr
aig
Glee
son,
Uni
vers
ity
Colle
ge Lo
ndon
(UK)
Shar
ing
stan
dard
ised
mod
els a
nd d
ata
on th
e O
pen
Sour
ce B
rain
pla
tform
(inv
ited
talk
)
11:5
0 - 1
2:10
Mel
anie
Ste
fan,
Uni
vers
ity o
f Ed
inbu
rgh
(UK)
Mul
ti-m
etho
d m
odel
ling
of sy
napt
ic pl
astic
ity a
nd
the
chal
leng
es it
brin
gs (i
nvite
d ta
lk)
12:1
0 - 1
2:20
Jörn
Sta
rruß,
TU
Dres
den
(D)
Prin
ciple
s for
dec
lara
tive
mul
ticel
lula
r mod
ellin
g
12:2
0 - 1
2:30
Guille
rm Y
anez
, Pon
tifici
a Un
iver
sidad
Cat
ólica
de
Chile
(CL)
Flap
jack
: an
open
-sou
rce
tool
for s
torin
g,
visu
alisi
ng,a
naly
sing
and
mod
ellin
g ki
netic
gen
e ex
pres
sion
data
12:3
0 - 1
2:45
all s
peak
ers o
f Ses
sion
14Di
scus
sion
12:4
5 –
13:0
0
Carl
Bosc
h Au
dito
rium
(H1)
Sem
antic
Mod
el A
nnot
atio
n; C
hair:
Dag
mar
Wal
tem
ath
(Gre
ifsw
ald,
D)
Emer
ging
Sta
ndar
diza
tion
Nee
ds a
nd M
ultic
ellu
lar M
odel
ing;
Cha
ir: T
im Jo
hann
(Dor
tmun
d, D
)
CLO
SIN
G PL
ENAR
Y; C
hairs
: Mar
tin G
oleb
iew
ski (
HITS
, Hei
delb
erg,
D) &
Dag
mar
Wal
tem
ath
(Uni
vers
ity o
f Gre
ifsw
ald,
D)
time
Plen
ary/
Brea
kout
sess
ion
1Br
eako
ut se
ssio
n 2
Brea
kout
sess
ion
3Ro
om:
Foye
rFr
itz H
aber
Aud
itoriu
m (H
2)Al
win
Mitt
asch
Stu
dio
(H3)
13:0
0 - 1
4:00
Lunc
h14
:00
– 15
:30
Sess
ion
15:
FAIR
DOM
PAL
s and
Use
r W
orks
hop;
Cha
ir: O
lga
Kreb
s (H
ITS,
Hei
delb
erg,
D)
EVER
YBO
DY IS
INVI
TED
TO A
TTEN
D:
ht
tps:
//fa
ir-do
m.o
rg/e
vent
s/fa
irdom
-pal
s-us
ers-
mee
ting-
2019
/
Mod
el e
Xcha
nge
cons
ortiu
m; C
hairs
: Hen
ning
He
rmja
kob
& R
ahum
an
Sher
iff (
EMBL
-EBI
, Hin
xton
, UK
)
15:3
0 - 1
6:00
Post
er S
essio
n &
Cof
fee
Brea
k16
:00
– 17
:30
Sess
ion
16:
FAIR
DOM
PAL
s and
Use
r W
orks
hop;
Cha
ir: O
lga
Kreb
s (H
ITS,
Hei
delb
erg,
D)
EVER
YBO
DY IS
INVI
TED
TO A
TTEN
D:
ht
tps:
//fa
ir-do
m.o
rg/e
vent
s/fa
irdom
-pal
s-us
ers-
mee
ting-
2019
/
17:3
0EN
D O
F M
EETI
NGCa
rl Bo
sch
Audi
toriu
m (H
1)
Keynote Talks
Computational Physiology and the Physiome Project July 1511:00amH1Peter Hunter1
1Auckland Bioengineering Institute, University of Auckland (New Zealand)
Computational physiological models deal with multiple physical processes (coupledtissue mechanics, electrical activity, fluid flow, etc) at multiple spatial and temporalscales. In many cases the goal is to understand integrative biological function interms of underlying tissue structure and molecular mechanisms. These models areintended both to help understand physiological function and to provide a basis fordiagnosing and treating pathologies in a clinical setting. The Physiome Project isdeveloping model and data encoding standards, web accessible databases and opensource software for multiscale modelling (e.g. www.cellml.org). A new journal calledPhysiome will soon be launched to publish reproducible and reusable models. Thetalk will discuss the current state of the Physiome Project, including a summary ofa multi-physics mathematical framework based on port-Hamiltonians for modellingphysiology, and a project to map the pathways of the autonomic nervous system(commonfund.nih.gov/sparc).
Implementing Open Science in publishing July 151:00pmH1Thomas Lemberger1
1EMBO Press, Heidelberg (Germany)
Computational physiological models deal with multiple physical processes (coupledtissue mechanics, electrical activity, fluid flow, etc) at multiple spatial and temporalscales. In many cases the goal is to understand integrative biological function interms of underlying tissue structure and molecular mechanisms. These models areintended both to help understand physiological function and to provide a basis fordiagnosing and treating pathologies in a clinical setting. The Physiome Project isdeveloping model and data encoding standards, web accessible databases and opensource software for multiscale modelling (e.g. www.cellml.org). A new journal calledPhysiome will soon be launched to publish reproducible and reusable models. Thetalk will discuss the current state of the Physiome Project, including a summary ofa multi-physics mathematical framework based on port-Hamiltonians for modellingphysiology, and a project to map the pathways of the autonomic nervous system(commonfund.nih.gov/sparc).
13
Modeling projects across platforms - the reality and howreality should beJuly 15
1:45pmH1 Ursula Kummer1
1Heidelberg University (Germany)
With the establishment of SBML many years ago a first and important step wastaken to be able to perform modeling projects across platforms. Thus it is nowpossible to set up a model using one software and perform specific analyses in othersoftware. However, despite the progress done, in reality there are still often trapsand boundaries one wouldn’t expect. In this talk examples from real world problemsare given and suggestions made how we could tackle the still existing problems.
Let’s go on a FAIR asset management safariJuly 189:30am
H1 Carole Goble1
1University of Manchester (UK)
FAIR: Findable Accessable Interoperable Reusable. The ”FAIR Principles” for re-search data, software, computational workflows, scripts, or any other kind of Re-search Object one can think of, is now a mantra; a method; a meme; a myth; amystery. FAIR is about supporting and tracking the flow and availability of dataacross research organisations and the portability and sustainability of processingmethods to enable transparent and reproducible results. All this is within the con-text of a bottom up society of collaborating (or burdened?) scientists, a top downcollective of compliance-focused funders and policy makers and an in-the-middleposse of e-infrastructure providers.
Making the FAIR principles a reality is tricky. They are aspirations not stan-dards. They are multi-dimensional and dependent on context such as the sensitivityand availability of the data and methods. We already see a jungle of projects, initia-tives and programmes wrestling with the challenges. FAIR efforts have particularlyfocused on the ”last mile” - ”FAIRifying” destination community archive reposi-tories and measuring their ”compliance” to FAIR metrics (or less controversially”indicators”). But what about FAIR at the first mile, at source and how do we helpAlice and Bob with their (secure) data management? If we tackle the FAIR firstand last mile, what about the FAIR middle? What about FAIR beyond just datalike exchanging and reusing pipelines for precision medicine?
Since 2008 the FAIRDOM collaboration [1] has worked on FAIR asset manage-ment and the development of a FAIR asset Commons for multi-partner researcherprojects [2], initially in the Systems Biology field. Since 2016 we have been work-ing with the BioCompute Object Partnership [3] on standardising computationalrecords of HTS precision medicine pipelines.
So, using our FAIRDOM and BioCompute Object binoculars let’s go on a FAIRsafari! Let’s peruse the ecosystem, observe the different herds and reflect what wherewe are for FAIR personalised medicine.
[1] http://www.fair-dom.org
14
[2] http://www.fairdomhub.org[3] http://www.biocomputeobject.org
From data via models to personalized medicine - Anexample in Pediatric Oncology July 18
11:45amH1Norbert Graf1
1Saarland University Medical Center, Homburg (Germany)
Medicine is undergoing a paradigm shift from phenotyping to genotyping. This issupported by systems approaches to disease, modeling and visualization technolo-gies, and new computational and mathematical tools. An open, modular architec-tural framework for tools, models and services is necessary to design and developmodels for translational medicine. Such a framework needs to share and handleefficiently the enormous personalized data sets to ensure that policies for privacy,non-discrimination, and access to data, services, tools and models are implementedto maximize data protection and data security to enable standardization and seman-tic data interoperability to integrate models from system biology to guarantee thattools, services and models are clinically driven and do enhance decision support toprovide tools for large-scale, privacy-preserving data mining, and literature miningto enhance patient empowerment. In addition, all models need to be evaluated andvalidated by end-users and have to be tested in concrete clinical research projectsthat target urgent topics of the medical community. To sustain such a translationalinfrastructure realistic use-cases have to offer tangible results for end-users. Teach-ing and educational programs should be implemented to facilitate the use of themodels. The large scale integrating transatlantic CHIC project (http://www.chic-vph.eu/) developed a suite of tools, services and a secure infrastructure that supportsaccessibility and reusability of mathematical and computational hypermodels andfulfills the above-mentioned criteria. One of these models is the Nephroblastoma(Wilms Tumor) Multimodeller Hypermodel. This model is answering the question,if preoperative chemotherapy will reduce the tumor size of a nephroblastoma in anindividual patient. For that purpose, different hypomodels are developed, includ-ing a vascular, metabolic, biomechanical, molecular model and the Wilms TumorOncosimulator. All of them compose the final multimodeler hypermodel. A clinicalResearch Application Framework (CRAF) allows a clinician to run the hypermodelin individual patients, by selecting the model and the data that are needed. Cur-rently we are starting to prospectively evaluate the model by comparing response tochemotherapy in reality with the predicted response by the hypermodel in individualpatients with nephroblastoma.
15
16
Invited Talks
Monday
The History of COMBINE 15 July10:20amH1Michael Hucka1
1Caltech, Pasadena (USA)
In this presentation, I will summarize COMBINE (the COmputational Modeling inBIology NEtwork), the history of its formation, and its status today.
How (not) to Apply (COMBINE) Standards inAgent-Based Modeling July 15
3:30pmH1Judith Wodke1, Jens Hahn1, Jorin Diemer1, Edda Klipp1
1Humboldt-Universitat zu Berlin, Theoretische Biophysik (Germany)
Agent-based modeling (ABM) has recently become of great interest in differentresearch areas, especially when modeling stochastic processes or when aiming atsingle molecule-resolution. The advantages of ABM are clear: i) modeling is cen-tered around the system entities, which can hold any properties the modeler is ableto give them, ii) (system) state transitions are defined by functions iii) the level ofdetail is limited by computer power only. However, this liberty in terms of pos-sibilities is paid for by a total lack of standardized methods and tools. There isno ’Copasi for ABM’, not a simple plotting function is pre-defined. Also, no listof species/reactions/parameters is created automatically. Even worse, parametersare often hidden in different files and at least hundreds of lines of code. Becauseno user interface (UI) exists that provides easy access to the model and its simu-lation options, accessibility and usability of agent-based models are highly limited.A modeler aiming at the application of modeling standards has to hard-code themwithin the model and/or its supplements (analysis and plotting tools, (G)UI, etc.).Thus, while the systems biology/modeling community successfully increased thelevel of standardization in mathematical modeling for years now, ABM poses a newchallenge to this noble aim. Usability of agent-based models and reproducibility ofresults from ABM simulations highly depend on capacities and disposition of theindividual modeler who not only has to put a significant effort in commenting thecode but needs to provide access and analysis tools along with the model. In sum-mary, while ABM allows us to model processes not describable by kinetic rate laws
17
only, it requires a significant amount of extra time and (wo)man power to apply aminimum of (COMBINE) modeling standards.
Computational modeling of liver function tests -Stratification and individualization based on semantic
annotations of models and dataJuly 154:00pm
H1 Matthias Konig1, Jan Grzegorzewski1
1Humboldt-Universitat zu Berlin (Germany)
Quantitative dynamical liver function tests evaluate the function of the liver via theclearance of a given test substance, thereby providing in vivo information on themetabolic capacity of the liver. Within this work we used multi-scale SBML models(core and comp package) for the description of whole-body physiological models ofabsorption, distribution, metabolization and elimination of test substances used indynamical liver function tests. Assessment of liver function is a key task in hepatol-ogy but accurate quantification of hepatic function has remained a clinical challenge.Dynamic liver function tests are a promising tool for the non-invasive evaluation ofliver function in vivo. One class of such tests are breath tests based on the conver-sion of 13C-labeled substrates by the liver to 13CO2 subsequently measured in thebreath. A commonly applied substrate is 13C-methacetin, converted to paraceta-mol and 13CO2 via cytochrome P450 1A2 (CYP1A2), used orally in the methacetinbreath test (MBT) and intravenously in the LiMAx test. An important clinicalquestion is which factors can affect MBT and LiMAx results. The aim of our studywas to answer this question using computational modeling to derive basic informa-tion for a better understanding of the methacetin breath test and factors influencingits results. The liver function test based on caffeine is known for long, but its clin-ical usability is hampered by large interindividual variability and dose-dependency.In this work we developed a detailed physiologically based pharmacokinetics model(PBPK) for the evaluation of the caffeine clearance test, assessing hepatic conversionof caffeine to paraxanthine via cytochrome P450 CYP1A2. The model is able toreproduce results from a wide range of reported studies under varying caffeine dosesand application routes. The model accounts for interindividual differences based ondistributions of CYP1A2 and modification of CYP1A2 activity via lifestyle factors,e.g. smoking, and pharmacological interactions, e.g. oral contraceptives. Valida-tion was performed with an independent clinical trial (EudraCT 2011-002291-16,ClinicalTrials.gov NCT01788254) demonstrating an improved prediction using indi-vidualized models accounting for smoking status and contraceptive use. Hereby, wecould reduce the large variability in the test results providing the basis for bettersensitivity and specificity in diagnosing subjects with liver problems.
18
Research software engineers and their role for open andreproducible research July 15
4:30pmH1Heidi Seibold1
1LMU Munich (Germany)
Many of the proposed solutions to the reproducibility crisis are technical solutions.Open and reproducible research require researchers to learn new technical skills. Inthis talk I will show some of the tools and strategies I use to make my researchopen and reproducible. I argue that not all researchers can become experts inthese tools and strategies. Instead we need research software engineers (RSEs) andreproducibility support.
19
Tuesday
A Standard Enabled Workflow for Synthetic BiologyJuly 169:30pm
H1 Chris Myers1
1University of Utah, Salt Lake City (USA)
A synthetic biology workflow is composed of data repositories that provide infor-mation about genetic parts, sequence-level design tools to compose these parts intocircuits, visualization tools to depict these designs, genetic design tools to selectparts to create systems, and modeling and simulation tools to evaluate alternativedesign choices. Data standards enable the ready exchange of information withinsuch a workflow, allowing repositories and tools to be connected from a diversityof sources. This talk describes one such workflow that utilizes the growing ecosys-tem of software tools that support the Synthetic Biology Open Language (SBOL)to describe genetic designs, and the mature ecosystem of tools that support theSystems Biology Markup Language (SBML) to model these designs. In particu-lar, this presentation will demonstrate a workflow using tools including SynBioHub,SBOLDesigner, and iBioSim. SynBioHub (http://synbiohub.org) is a database de-signed for storing synthetic biology designs captured using the SBOL data model,and it provides both a RESTful API for computational access and a user-friendlyWeb-based frontend. SBOLDesigner is a sequence editor that allows the designerto fetch parts from a SynBioHub repository and compose them to construct largerdesigns. Finally, iBioSim is genetic modeling, analysis, and design tool that pro-vides a means to construct SBML models for these designs that can be simulatedand analyzed using a variety of techniques. Both SBOLDesigner and iBioSim alsosupport uploading these larger system designs back to the SynBioHub repository.Finally, this talk will demonstrate how this workflow can be utilized to produce acomplete record of a genetic design facilitating reproducibility and reuse.
BIOROBOOST: A last chance for standardisation inBiology?July 16
10:00amH1 Manuel Porcar Miralles1
1University of Valencia (Spain)
Synthetic Biology is an engineering research field aiming at (re)designing biologicalcircuits for applied purposes. SynBio strongly relies on the use of well-defined, uni-versal and robust standard components. Despite the outstanding success of syntheticbiology in the last years, the difficulties in defining biological standards are wellknown. Some of the difficulties to achieve standardisation include the cultural ”trib-alisation” of synthetic biology, a field involving mainly biologists/biotechnologistsand engineers; the intrinsic features of live such as mutation, noise, metabolicpromiscuity, emergent properties, fitness biasses, variability and, finally, evolution.BIOROBOOST proposes to finally overcome cultural issues and to dramatically ad-vance in solving technical difficulties by i) gathering the most relevant stakeholders
20
of all the aspects of standardisation in biology in a co-creation scenario; ii) by em-pirically testing cultural (lab-centric) standardisation practices and by promoting aconsensus conceptual and technical redefinition of biological standards; and, finally,iii) by fostering a realistic and flexible toolbox of standard biological parts, includ-ing a reduced set of specialised chassis for specific applications as well as a renewedconceptual framework to inform policy makers, scientific and other societal actors.
The Center for Reproducible Biomedical Modeling July 1611:30amH1Herbert Sauro1,Michael L. Blinov, John H. Gennari, Arthur P. Goldberg,
Jonathan R. Karr, Ion I.Moraru, David P. Nickerson1University of Washington, Seattle (USA)
The Center for Reproducible Biomedical Modeling is a NIH-funded center whichaims to enable comprehensive predictive models of biological systems, such aswhole-cell models, that can guide medicine and bioengineering. Achieving this goal requiresnew tools, resources, and best practices for systematically, scalably, and collabora-tively building, simulating and applying models, as well as new researchers trained incomprehensive modeling. To meet theseneeds, the center is developing new technolo-gies for comprehensive modeling, working with journals to provide authors, review-ers, and editors model annotation and validation services, and organizing coursesand meetings totrain researchers to model systematically, scalably, and collabora-tively. For more information see: https://reproduciblebiomodels.org/. The talk willinclude the latest developments from the centers efforts.
21
Wednesday
A personalized modeling pipeline for cardiacelectrophysiology simulations of cardiac resynchronization
therapy in infarct patientsJuly 179:30am
H1 Caroline Mendonca Costa1, Aurel Neic2, Eric Kerfoot1, Bradley Porter1,Benjamin Sieniewicz1, Justin Gould1, Baldeep Sidhu1, Zhong Chen1, GernotPlank2, Christopher A. Rinaldi1,3, Martin J. Bishop1, Steven A. Niederer1
1King’s College London (UK), 2Medical University of Graz (Austria), 3Guy’s and St.Thomas’ Hospital, London (UK)
Cardiac Resynchronization Therapy (CRT) is an effective treatment for heart fail-ure. However, CRT increases the risk of ventricular tachycardia (VT) in patientswith ischemic cardiomyopathy (ICM) when the left ventricular (LV) epicardial leadis implanted in proximity to scar. To determine the mechanisms underpinning thisrisk, we investigate the effects of pacing on local electrophysiology (EP) in rela-tion to scar that provides a substrate for VT in ICM patients undergoing CRT.Investigating the role of pacing location during CRT on VT risk in the clinical set-ting is hampered by the anatomy of the coronary sinus veins, which limits whereCRT pacing leads can be implanted. Computational models provide full flexibility,thus, allowing a detailed and systematic investigation of the role of pacing location.However, building clinically relevant models that include realistic anatomy and scarmorphology, and simulating cardiac electrophysiology (EP) at clinical time-scales re-mains a challenging task. Therefore, we developed a personalization pipeline, whichallows fast development of infarct model cohorts and EP simulations. The devel-oped pipeline was used to build 24 image-based models of left ventricular anatomyand scar morphology. The pipeline allows fast ( 2 hours) creation of models in astandardized fashion. Moreover, the developed pipeline allows selecting lead loca-tions consistently across models, improving control of simulation conditions. Themodels built were used to investigate the role of pacing location on dispersion of re-polarization, as a surrogate for VT risk. Our simulation results predict that pacingadjacent to a scar increases dispersion of repolarization in its vicinity, which pro-vides a mechanistic explanation for increased VT risk in CRT patients with infarct.Moreover, our simulations predict that pacing 3.5 cm or more from a scar is likelyto avoid increasing VT risk in CRT patients.
22
CellDesigner: A modeling tool for biochemical networks July 1711:15amH1Akira Funahashi1
1Keio University (Japan)
Understanding the logic and dynamics of gene-regulatory and biochemical networksis a significant challenge of systems biology. To facilitate this research topic, wehave developed a modeling/simulating tool called CellDesigner.
CellDesigner primarily has capabilities to visualize, model, and simulate gene-regulatory and biochemical networks. Two significant characteristics embedded inCellDesigner boost its usability to create /import/export models: (1) Solidly definedand comprehensive graphical representation (Process Diagram) of network modelsand (2) Systems Biology Markup Language (SBML) as a model-describing basis,which function as inter-tool media to import/export SBML-based models.
Since its initial release in 2004, we have extended various capabilities of CellDe-signer. For example, we integrated the third party Garuda enabled simulation/analysissoftware packages. CellDesigner also supports simulation and parameter scan, sup-ported by integration with SBML ODE Solver, COPASI and Simulation Core Li-brary enabling users to simulate through our sophisticated graphical user interface.Users can also browse and modify existing models by referring to existing databases(ex. BioModels.net) directly through CellDesigner.
Those extended functions empower CellDesigner as not only a modeling/simulatingtool but also an integrated analysis suite. CellDesigner is implemented in Java andthus supports various platforms (i.e., Windows, Linux, and MacOS X). CellDesignerversion 4.4.2 is freely available via our web site http://celldesigner.org, and a newversion of CellDesigner is under development, which will support spatial modeling.
Virtual Cell: Modeling and Visualization of Reaction Rules July 1711:45amH1Michael Blinov1
1UConn School of Medicine, Farmington (USA)
Rule-based approach allows representation and simulation of biological systems ac-counting for molecular features (molecular sites of binding and modifications) andsite-specific details of molecular interactions. While rule-based description is veryprecise and can define very fine molecular details (like how phosphorylation statusof a single residue in a multi-protein complex can affect affinity of another bindingsite of another protein within the same complex), it comes with a cost. Combiningall the assumptions scribed in multiple rules into a single diagram is a dauntingtask. Here we present the approach and software for precise and compact represen-tation of rule-based models implemented in Virtual Cell modeling and simulationtool. It is based on the three basic concepts that allow scalability: molecular pattern(molecules that participate or affect the rule), rule center (molecular sites that aredirectly modified by a rule), and rule context (molecular sites that affect the rule).We also provide an approach for a complete visualization of rule-based modeling
23
in Systems Biology Graphical Notations (SBGN)-compliant Process Diagram (PD)conventions, suggesting a possible SBGN-PD extension for rule-based models.
24
Thursday
EU-STANDS4PM: A pan-European Expert Forum joinedforces to tackle the complexity of big data integration for in
silico methodologies in personalised medicine July 1811:30amH1Marc Kirschner1
1Forschungszentrum Julich (Germany)
On January 1st 2019 the Horizon2020 Coordinating and Support Action EU-STANDS4PM - A European standardization framework for data integration and data-drivenin silico models for personalized medicine launched activities with support by theEuropean Commission Directorate-General for Research & Innovation. During thenext three years EU-STANDS4PM will initiate an EU-wide mapping process to as-sess and evaluate strategies for data-driven in silico methodologies for personalisedmedicine. This process is the foundation to assemble specific recommendation andguidelines (including formal standard documents) for data harmonization and in-tegration strategies as well as data-driven in silico approaches to interpret humandisease and health data. EU-STANDS4PM is an open network and seeks input fromall relevant stakeholders that have an interest in advancing in silico methodologies inpersonalised medicine through broadly applicable standards as well as coordinatedprocedures for integration and harmonization of heterogeneous health data. Thiswill help to sustain the competitiveness of the European Research Area and ensurea leading role for the European personalized medicine community of stakeholders inthe transition from current reactive medical practice to a data-driven and predic-tive medicine of the future. EU-STANDS4PM receives funding from the EuropeanCommission under grant agreement 825843.
Standards in Biobanking July 1812:15pmH1Petr Holub1
1BBMRI-ERIC (Austria) and Masaryk University (Czech Republic)
Standardization efforts for Biobanking will be discussed, including activities of theEuropean research infrastructure for biobanking BBMRI-ERIC (Biobanking andBiomolecular Resources Research Infrastructure European Research InfrastructureConsortium) and standardization efforts within the technical committee for biotech-nology standards of the international standardization organization ISO (ISO/TC 276Biotechnology).
25
Standards for Clinical Data and Systems MedicineJuly 1812:30pm
H1 Søren Brunak1
1University of Copenhagen (Denmark)
Standardization efforts for clinical data will be discussed, especially with regard topreparing the data for their use in systems medicine and computer modelling in thecontext of personalized medicine.
A Bird’s Eye view of Legal and Ethical aspects ofEU-STANDS4PM18 July
12:45pmH1 Katharina O Cathaoir1
1University of Copenhagen (Denmark)
This presentation will introduce the work-in-progress of Work Package 3: legal andethical framework for data-driven in silico models in personalized medicine. Theoverarching aim of this work package is to provide comprehensive guidance on thelegal, ethical and policy considerations arising from the use of in silico modellingfor personalised medicine through mapping overlapping but distinct regulations andconventions: the European Union General Data Protection Regulation (GDPR)(2016), the EU Clinical Trials Regulation (2014) and Council of Europe conventionson patients rights, such as the Convention on Human Rights and Biomedicine (1997).The purpose of the presentation is to highlight the legal significance of the source ofthe data used in in silico modelling. For example, certain legal and ethical principlesapply to data gathered in the doctors office, while other rules apply to data gatheredin connection with clinical trials. From a GDPR perspective, the basis upon whichthe data was originally processed can be significant, for example, whether consentwas given, and if so, what form of consent. At international and European levelspatient and research subjects rights are underpinned by separate principles, yet thissystem of governance is weaker, in light of more modest means of enforcement, whichwill be outlined. Finally, one can also question the extent to which these frameworksremain applicable to Big Data research.
26
In silico medicine and the VPH Institute: bringing thecommunity together and the field forward 18 July
14:00pmH1Liesbet Geris1
1VPH Institute (Belgium)
Nearly a decade after its official incorporation, the VPH institute, a scientific or-ganization by and for researchers working the field of in silico medicine, is gettingevery closer to fulfilling its mission: ensuring the adoption of in silico technologiesin all aspects of healthcare, from prevention to clinical translation. Hereto, theinstitute works together with a wide variety of stakeholders, from researchers toclinicians, from industry to the regulators, and from policy makers to funders. Inrecent years, the field of in silico medicine has gotten a serious boost owing to anumber of activities at the translational side. For instance, in 2016, FDA publishedguidelines on reporting of computer models used in the context of medical devices.More recently, in 2018, ASME published a standard on verification and validationof computer models, again in the context of medical devices, but quite generic interms of the underlying philosophy. Guidelines and standards such as these do notonly reinforce the confidence of industrial and clinical stakeholders in the potentialof in silico models, they should also be adopted by the research community in orderto increase the quality and reproducibility of the scientific work that is published.
27
Friday
Semantic annotation in the Physiome Model RepositoryJuly 199:30am
H1 David Nickerson1, Dewan Sarwar1, Tommy Yu1, Anand Rampadarath1, KorayAtalag1
1Auckland Bioengineering Institute, University of Auckland (New Zealand)
The Physiome Model Repository (PMR) contains over 800 publicly available andversion controlled workspaces. These primarily consist of published computationalmodels encoded in the CellML format, some including descriptions of simulationexperiments encoded in the SED-ML format. Here I will briefly introduce our recentefforts to curate and annotate these models with detailed biological semantics aswell as some high-level workspace annotations. I will then demonstrate how theseannotations can be utilised to improve the discovery and comprehension of modelsin the PMR and then how they aid the model building tasks. I will conclude withsome thoughts on how models built in this manner can be verified against similarlyannotated simulation experiments or experimental data.
Gene Ontology Causal Activity Modeling (GO-CAM)July 1910:00am
H1 Huaiyu Mi1, Paul D. Thomas, David P. Hill, David Osumi-Sutherland, KimberlyVan Auken, Seth Carbon, James P. Balhoff, Laurent-Philippe Albou, Benjamin
Good, Pascale Gaudet, Nomi L. Harris, Suzanna E. Lewis, Christopher J. Mungall1University of Southern California (USA)
The Gene Ontology was developed to conceptualize and describe the functions ofgenes. The broad aims were (1) comprehensive representation of how genes functionin biological systems to enable computational analysis of genomic datasets, and (2)representation of biological concepts consistently across the broad range of modelorganisms to enable the identification of evolutionarily shared genetic programs, andto use this information to shed light on the functions of human genes. GO anno-tations are regarded as the most comprehensive structured representation of genefunctions, and are widely used in the interpretation of genome-wide experimentaldata. However, the current GO annotation data structure allows association ofGO terms to individual gene product, so it limits the expressiveness of annotationsand their application in computational analysis of experimental data. To addressthis limitation, the GO Consortium has developed a framework, GO Causal Activ-ity Modeling (GO-CAM), for linking multiple GO annotations into an integratedmodel of a biological system. The fundamental framework of GO-CAM is to connectmolecular functions associated to different gene products with causal relationships.GO-CAM supports modeling at multiple levels, from individual gene products tocomplex regulatory and metabolic pathways, and can be applied in network analysisand systems biology modeling. In addition, GO-CAM is actively working with otherstandards (BioPAX/SBGN) and resources (Reactome) with regard to interoperabil-ity and compatibility.
28
Sharing standardised models and data on the Open SourceBrain platform July 19
11:30amH1Padraig Gleeson1
1University College London (UK)
I will present the latest functionality of the Open Source Brain platform (OSB;http://www.opensourcebrain.org). OSB was created for sharing and collaborativelydeveloping models in computational neuroscience. Models of cells and circuits instandardised formats (NeuroML) can be visualised, analysed and simulated througha standard web browser. The aim has been to improve the quality, accessibility andscientific rigour of models used to investigate brain function. Following a renewalof our funding from the Wellcome Trust, we are expanding the functionality ofthe platform to also enable sharing of the experimental data behind the models.I will present how we are using the Neurodata Without Borders (NWB) standardto facilitate sharing of cellular neuroscience data and will outline our vision for aplatform of complex yet accessible neuronal models which are shared along with theexperimental recordings used to constrain them.
Multi-method modelling of synaptic plasticity and thechallenges it brings July 19
11:50amH1Melanie Stefan1
1University of Edinburgh (UK)
We can look at processes underlying learning and memory at a variety of differentscales: from investigating molecular interactions within synapses to studying net-works of neurons, to looking at how students learn in classrooms. Computationalmodelling has proven to be a promising tool at all those levels. But an increasinglyimportant question is how those levels interact with each other and how we can com-bine different types of models as well as experimental data to answer this question.Addressing this question requires thinking about interfaces both between levels ofanalysis, and between different methods. I will present some of our preliminary workand future directions in this area.
29
30
Contributed Talks
Tuesday
Analyzing Genetic Circuits for Hazards and Glitches July 1610:30amH1Pedro Fontanarrosa1, Hamid Hosseini2; Amin Borujeni2, Yuval Dorfan2, Chris
Voigt2, Chris Myers1
1University of Utah (USA), 2MIT, Boston (USA)
A hazard is the possibility of an unwanted or unexpected variation of the output ofa combinational logic network before it reaches steady-state. A glitch is the actualobservance of such a problem. These terms are used mostly for electronic circuits,though glitches have been observed for genetic regulatory networks (GRNs) as well.A glitch is a transient behavior that corrects itself as the system reaches a steady-state. Nonetheless, this glitching behavior can have drastic consequences if thistransient output of the GRN causes an irreversible change in the cell such as acascade of responses within or with other cells, or if it induces apoptosis. Therefore,avoiding glitching behavior can be crucial for safe operation of a genetic circuit.Tobetter understand glitching behavior in genetic circuits, this work utilizes a versionof our dynamic model generator that automatically generates a mathematical modelcomposed of a set of ordinary differential equations (ODEs) that are parameterizedusing characterization data found in a Cello genetic gate library. Simulation of adynamic model allows for the prediction of glitches that cannot be observed withsteady-state analysis. This work is done using data-standards such as SBOL, SBMLand SED-ML. For more information visit: https://journal.physiomeproject.org
SBOL and its applicability in partially and fully automateddesign workflows: Three success stories July 16
10:40amH1Ernst Oberortner1
1DOE Joint Genome Institute (JGI), Berkeley Labs (USA)
The Synthetic Biology Open Language (SBOL) and its recently added W3CProvenance Ontology (PROV-O) extension enable to exchange data in a standard-ized format and to track the activities, entities and data across the entire, iterativelife cycles of synthetic biology Design-Build-Test-Learn (DBTL) workflows. An in-tegrated design workflow is required when managing various large-scale, complexprojects simultaneously, as we experience at the U.S. Department of Energy (DOE)
31
Joint Genome Institute (JGI). The dilemma is, however, to address the varying re-quirements across the simultaneous projects, to define the activities of each projectsDBTL workflow, and to incorporate frequently changing requirements due to im-proved technologies and scientific discoveries. In this presentation, I will survey thedesign of three projects with varying requirements and workflows. The projects fo-cus on the design of a few small proteins to pathways with multiple operons, such asin refactored gene clusters. The design workflow was either performed using a par-tially or fully automated approach including the adoption of SBOL for exchangingdesigns and tracking design activities. The design tasks of the projects range from(i) codon optimizing the genes for expression in the target host organism over (ii)grouping genes into operons to modulate their expression using regulatory elements(e.g., promoters, terminators) to (iii) specifying the synthesis limitations as well asthe assembly and cloning instructions of the design. When designing a pathway, thenvarious design alternatives need to be considered. We experienced designs rangingfrom explicitly specified pathway variants to combinatorial designs that need to besampled either randomly, algorithmically or based on constraints. Lastly, I will alsotouch on our approach of using SBOL for the specification of build instructions(i.e., synthesis, assembly, cloning) for the three projects. Although in its infancy,this approach could enable to order the physical construction of a design and de-sired build instructions to downstream manufacturing facilities, such biofoundriesor commercial DNA synthesis providers.
Physiome - Publish your models curated for reproducibilityand reusability to increase research quality for everyoneJuly 16
12:00pmH1 Karin Lundengard1, Peter Hunter1, David Nickerson1
1Auckland Bioengineering Institute, University of Auckland (New Zealand)
Physiome is a journal committed to reproducibility and reusability of mathe-matical models of physiological processes. Every article published in Physiome isconnected to a curated and permanent version of the model code with a persistentidentifier. Through the Physiome paper, the code necessary to run the model iseasily accessible by just clicking a link, to be reused as it is or as a module in abigger model. It is also connected to a primary paper published in a domain spe-cific journal, where the validation and scientific value of the model is discussed.A Physiome publication is a complement to your primary article that ensures re-producibility, reusability and discoverability of you model. The format encouragesmodularity that facilitates combination of different models to develop the next levelof systems understanding. And all the models are in one place, easy to find andaccessible. Reproducibility and confirmation of results is crucial for useful scienceand should be one of the supporting pillars of good research. Yet, publication ofit is rarely incentivised, often treated as a secondary result at best, which under-mines the quality of our work. With the strict formulation of equations and easilyshared code, it seems like mathematical models should be reproducible by default,but in fact less than 10% of the models published in scientific journals work whenimplemented by another group. Waste no more valuable time and effort on trying
32
to implement models from papers that lack information, or having your results lostbecause others cannot use them. Publish your models in Physiome and contributeto making science useful in society (and less frustrating for your colleagues). Formore information visit: https://journal.physiomeproject.org
Reproducibility in model construction, validation andanalysis workflows in systems biology projects; Xylose
metabolism in Caulobacter crescentus as a case study, usingJWS Online and the FAIRDOMHub July 16
12:10pmH1Jacky Snoep1, Lu Shen2, Bertina Siebers2, Dawie van Niekerk1, Wolfgang
Muller3, Carole Goble4
1Stellenbosch University (South Africa), 2University Duisburg-Essen (Germany),3Heidelberg Institute for Theoretical Studies (HITS) (Germany), 4University of
Manchester (UK)
With the uptake of modelling standards as advocated by the COMBINE community,the description and publication of models in standard formats such as SBML hasmuch improved. Many journals require authors to make their mathematical modelsavailable, and the SBML format is often used in Systems Biology studies. However,the requirement to submit a model description does not guarantee that the modelsimulations shown in the manuscript are reproducible. Usually, reviewers will notcheck the submitted models, and without clear instructions for simulations, it canbe quite challenging to reproduce particular model simulations. Having been activefor a long time in technical model curation of scientific manuscripts that containmathematical models, we need on average about 5 communications with authorsto successfuly reproduce their simulation results, and model simulations publishedin journals without a curation service will very often not be reproducible. Usinga standard model simulation description such as SED-ML makes it much easier tocheck for reproducibility, but not many manuscripts include SED-ML descriptions ofmodel simulations. The JWS Online project includes both a model database and amodel simulation database, the tools to construct SBML and SED-ML files, and testfor model simulation reproducibility. In addition, it allows for linking experimentaldata stored on the FAIRDOMHub. Whereas one could consider the reproducibilityof model simulations as a minimal requirement, our aims for the scientific processof mathematical modelling should be much higher, including transparancy and re-producibility for the construction, validation and analysis of the model. We willillustrate the approach for a combined experimental and modelling study on xylosemetabolism, using initial rate kinetics on isolated enzymes, progress curve analysis,sequential enzyme cascade, one pot cascade and cell extract incubations. All data,model and simulation description files for each of the steps are made available onthe FAIRDOMHub, making the complete process reproducible and FAIR.
33
Causal, Mechanistic Pathway Based Analysis of -OmicProfilesJuly 16
12:20pmH1 Emek Demir1, Ozgun Babur, Funda Durupinar Babur, Hannah Manning, Michal
Grzadkowski, Joseph Estabrook, Olga Nikolova, Kevin Watanabe-Smith1Oregon Health & Science University (USA)
Genomics and imaging data produced by large NIH projects now reaches to Exabytescale. Current mechanisms of knowledge representation and scientific communica-tion in biology cannot adequately deal with the complexity and volume of thisinformationa serious bottleneck for developing a causal, predictive understanding ofthe cell. To address this need we have developed multiple algorithms and tools overthe years that bridges causal mechanistic knowledge obtained through decades oflow-throughput biology research with big data. Over the last two years these toolsare being actively used for molecular tumor boards and cancer precision medicinewithin the OHSU’s SMMART program. In this talk, we will describe the currentanalysis workflows, how they are facilitated by BioPAX and SBGN, initial tumorboard results as well as our future plans.
Predictive in-silico multiscale analytics to support cancerpersonalized diagnosis and prognosis, empowered by
imaging biomarkersJuly 1612:30pm
H1 Polyxeni Gkontra1, Leonor Cerd Alberich1, Gracia Mart Besa1, Luis MartBonmat1
1La Fe Health Research Institute (IIS La Fe), Valencia (Spain)
PRIMAGE (PRedictive In-silico Multiscale Analytics to support cancer personal-ized diaGnosis and prognosis, Empowered by imaging biomarkers) is a novel, highlyinnovative H2020 project (GA- 826494) aiming at providing cloud-based computa-tional solutions to the diagnosis, prognosis, choice and follow-up of treatment fortwo of the most common paediatric cancers with high societal impact, neuroblas-toma (NB) and Diffuse Intrinsic Pontine Glioma (DIPG). NB is the most frequentsolid cancer in childhood, while DIPG is the leading cause of brain tumour-relateddeath in children. Due to the highly complex nature of both tumours, their diag-nosis, prognosis and monitoring require the combination of several sources of datasuch as biological, clinical, multi-omics and imaging. Apart from the data itself,adequate data infrastructures ensuring secure and anonymized data management,storage and sharing, as well as computational approaches for the effective data ex-ploitation are of paramount importance. Particularly in the case of imaging data,automated high-throughput quantitative image analysis approaches (radiomics) andartificial intelligence are essential in order to translate the huge amount of highlycomplex images acquired by todays imaging systems into quantitative information.This information can be used to identify novel imaging biomarkers for disease di-agnosis, monitoring of progression, choice and response to therapy. In this context,PRIMAGE will develop a functional prototype cloud-based platform offering predic-tive tools to assist management of NB and DIPG. Multi-modal real-word imaging
34
data (MRI, CT, MIBG/SPECT, PET-FDG if MIBG is negative) from hospitalsaround Europe will be used to identify and validate reliable and reproducible imag-ing biomarkers for NB and DIPG using cutting-edge technologies. The data in-frastructures, imaging biomarkers identified, as well as the computational solutionsdeveloped within PRIMAGE, including models for in-silico medicine research, willbe validated in the context of NB and DIPG, but their application is not limited inthese types of cancers but can be extended in a large variety of cancers.
35
Wednesday
Datanator: Tools for Aggregating Data for Large-ScaleBiomodelingJuly 17
10:00amH1 Yosef Roth1, Zhouyang Lian1, Saahith Pochiraju1, Jonathan Karr1
1Icahn School of Medicine at Mount Sinai (USA)
Systems biology aims to understand how genotype influences phenotype. Com-prehensive mechanistic models, such as whole-cell models, can be used to explorecell dynamics. However, the data required to construct these models are containedacross many different databases – often with inconsistent identifiers and formats.In addition, the time required to manually collect this data impedes the creationof large-scale models. To accelerate whole-cell modeling, we developed Datanator,an integrated database and search engine. Datanator allows a modeler to inputa desired biological parameter (e.g. molecular concentration, reaction rate, etc.)with desired search criteria (e.g. organism, environmental conditions, etc.), andDatanator returns a list of the most relevant observations together with a singleconsensus value that can be directly used in a model. To accomplish this, Datana-tor integrates genomic data (RefSeq, Ensembl), transcriptomic data (ArrayExpress),proteomic data (Pax-DB), kinetic data (SABIO-RK), and metabolite concentrationdata (ECMDB, YMDB). Datanators search engine can identify the most relevantdata from a combination of search criteria: genetic similarity (KEGG orthology),taxonomic similarity (NCBI taxonomy), molecular similarity (tanitomo string com-parison), reaction similarity (EC number), and environmental similarity. Datanatorenable scaleable model creation and consistent provenance tracking. In addition tousing Datanator to build a WC model of Mycoplasma pneumoniae, we have shownthat Datanator can find missing parameters for ODE models, augment FBA mod-els with kinetic bounds, and recalibrate models to similar organisms. Availability:https://github.com/KarrLab/datanator
Simulation and Sensitivity Analysis for Large KineticModels in AMICIJuly 17
10:10amH1 Fabian Frohlich1
1Harvard Medical School (USA)
In recent years, large, kinetic, multi-pathway models with hundreds to thousandsof species and parameters have become increasingly abundant. These large kineticsmodels are often constructed with the aim to deepen our mechanistic understandingof signaling pathways. Yet, calibration of these models poses a major computa-tional challenge, which limits our ability to work with these models in practice.Gradient-based methods such as quasi-Newton optimization or Hamiltonian Monte-Carlo have been demonstrated to work well for high-dimensional problems, butrequire methods to robustly compute parameter sensitivities. To address this issue,we developed AMICI (Advanced Multilanguage Interface to CVODES and IDAS),
36
a C++ library tailored to the simulation and sensitivity analysis of large OrdinaryDifferential Equation and Differential Algebraic Equation models. AMICI providespython and MATLAB interfaces that allow compilation of simulation executablesfrom SBML and BNGL formats. For gradient computation, AMICI implementssymbolic processing of model equations to provide efficient local sensitivity analysisusing forward, adjoint and steady-state sensitivity analysis. So far AMICI has beenapplied for simulation and model calibration in 27 publications. To highlight AM-ICI features that are essential for model simulation and sensitivity analysis for largemodels, I will showcase several examples, which includes multi-pathway signalingmodels with hundreds to thousands of species, reactions and parameters.
OpenCOR: current status and future plans July 1710:20amH1Alan Garny1, David Brooks1, Peter Hunter1
1University of Auckland (New Zealand)
OpenCOR: current status and future plans”,”OpenCOR is an open source cross-platform modelling environment. It can be used to organise, edit, simulate andanalyse models encoded in the CellML format. Partial support for SED-ML andCOMBINE archives is also available. In this talk, we will give an overview ofOpenCOR and illustrate how it fits within the overall COMBINE effort. We willthen share some of our future plans for OpenCOR.
Visualization of Part Use in SynBioHub July 1712:15pmH1Jeanet Mante1, Zach Zundel1, Chris Myers1
1University of Utah (USA)
We present a visualization plugin to assist designers in finding components for theirdesigns from SynbioHub instances. In particular, our plugin displays up to fourgraphs per part page. First, there is a Sankey diagram that shows other componentsthat are commonly used with this component sort by their type. Second, there isanother Sankey diagram that indicates which of these parts commonly come beforeor after the component of interest. Third, there is a histogram that shows howcommonly used this component is relative to most commonly used components.Finally, there is a histogram that shows the commonality of this component relativeto other commonly used components of the same type.
37
PySB framework: Tools to build, calibrate and visualizebiochemical modelsJuly 17
12:25pmH1 Oscar O. Ortega1, Carlos F. Lopez1
1Vanderbilt University (USA)
Computational models are used to understand the dynamics and behavior of com-plex biological network processes. In order to study cellular network processes,mechanistic models need to be built, calibrated to experimental data, and analyzedto generate hypotheses about the mechanisms that control different cellular pro-cesses. Here we present the PySB modeling framework of tools to build, calibrateand visualize models. PySB is a Python-based programming framework for systemsbiology, and it builds on BioNetGen and Kappa rule-based languages as well asPython libraries to enable model definition as functions that represent biologicalprocesses. PySB comprises three different tools for model calibration to experimen-tal data: SimplePso uses the Particle Swarm Optimization algorithm, PyDREAMuses the Differential Evolution Adaptive Metropolis (DREAM) algorithm to obtainposterior distributions of calibrated parameters, and Gleipnir that uses nested sam-pling algorithms for Bayesian parameter inference. Additionally, PySB containsa tool to obtain static and dynamic representations of model networks and theirsimulated dynamics. Importantly, the entire PySB framework can be employedwithin Jupyter Notebooks thus facilitating reproducibility and shareability of com-plete analysis pipelines. Therefore, we believe that the PySB framework enablesusers to obtain important mechanistic insights about complex biological processesthat could be potentially used for the development of novel therapies to treat dif-ferent human diseases.
Visualization, Access, and Exploration of BiologicalPathway Information from Pathway CommonsJuly 17
12:35pmH1 Augustin Luna1, Emek Demir2, Igor Rodchenkov3, Ozgun Babur2, Jeffrey
Wong3, Chris Sander1, Gary Bader3
1Dana-Farber Cancer Institute/Harvard Medical School (USA), 2Oregon Health &Science University (USA), 3University of Toronto (Canada)
Pathway Commons (pathwaycommons.org) serves researchers by integrating datafrom public pathway and interaction databases and disseminating this data in auniform fashion. The knowledge base is comprised of metabolic pathways, geneticinteractions, gene regulatory networks and physical interactions involving proteins,nucleic acids, small molecules, and drugs. Alongside attempts to increase the scopeand types of data, a major focus has been the creation of user-focused tools, re-sources, and tutorials that facilitate access, discovery, and application of existingpathway information to aid day-to-day activities of biological researchers. PathwayCommons offers a number of tools for accessing and searching the integrated datasetsthat are provided as file downloads in the Biological Pathway Exchange (BioPAX),Simple Interaction Format (SIF) and gene set (GMT) formats. This data is also pro-vided via web services that allow for integration with external tools (e.g., CyPath2,
38
a Cytoscape app, and the paxtoolsr R package; each is a network analysis tool).Discussion of Pathway Commons will, in part, focus on reusable web applicationvisualization components that are built upon cytoscape.js and the Systems BiologyGraphical Notation Markup Language (SBGN-ML). The components are the ba-sis for our web-based ’Search’ application that enables users to query pathways bykeyword and visualize returned pathways using SBGN with an automated layout.Additionally, these components are used for the ongoing development of variouspathway visualization applications, such as the online Newt SBGN pathway editor(newteditor.org). Overall, these software components enhance the accessibility topathways from third-party applications wishing to integrate support for pathwayvisualization and interpretation.
39
Thursday
ELIXIR Cloud & AAI: Standardised and InteroperableServices for Human Data CommunitiesJuly 18
10:15amH1 Susheel Varma1, Shubham Kapoor1, Ania Niewielska1, Alexander Kanitz1, Jinny
Chien, Andrea Cristofori, Sarah Butcher, Foivos Gypas, Kevin Sayers1, JuhaTornroos, Marco Tangaro, Giacinto Donvito, Yasset Perez-Riverol, Bjorn Gruning,
Jennifer Harrow, Jonathan Tedds, Salvador Capella, Tommi Nyronen, StevenNewhouse
1European Bioinformatics Institute (EMBL-EBI) (UK)
Currently, patient data are geographically dispersed, difficult to access, and often,patient data are stored in siloed project-specific databases preventing large-scaledata aggregation, standardisation, integration/harmonisation and advanced diseasemodelling. The ELIXIR Cloud and Authentication & Authorisation Infrastructure(AAI) for Human Data Communities project aim to leverage a coordinated net-work of ELIXIR Nodes to deliver a Global Alliance for Genomic Health (GA4GH)standards-compliant federated environment to enable population scale genomic andphenotypic data analysis across international boundaries and a potential infrastruc-ture to enable 1M Genome analysis. The ELIXIR Cloud & AAI project will laythe groundwork to deliver the foundational capability of federation of identities,sensitive data access, trusted hybrid cloud providers and sensitive data analysis ser-vices across ELIXIR Nodes by underpinning the bi-directional conversation betweenpartners with the GA4GH standards and specifications and ELIXIR trans-nationalexpertise. The project is also developing a framework for secure access and analysisof sensitive human data based on national federations and standardised discoveryprotocols. The secure authentication and authorisation process alongside guidelinesand compliance processes is essential to enable the community to use these datawithout compromising privacy and informed consent. The project therefore providesa globally available curated repository to store bioinformatics software containersand workflows (Biocontainers - GA4GH TRS), a service to discover and resolve thelocations of datasets (RDSDS - GA4GH DRS) and distributed workflow and taskexecution service (WES-ELIXIR/TESK - GA4GH WES/TES) to leverage the feder-ated life-science infrastructure of ELIXIR. The ambition of the project is to providea global ecosystem of joint sensitive data access and analysis services where feder-ated resources for life science data are used by national and international projectsacross all life science disciplines, with widespread support for standard componentssecuring their long-term sustainability. Connecting distributed datasets via commonstandards will allow researchers unprecedented opportunities to detect rare signalsin complex datasets and lay the ground for the widespread application of advanceddata analysis methods in the life sciences.
40
BioModels Parameters: A resource to search and accessparameters from published systems models July 18
10:25amH1Sheriff Rahuman1, Chinmay Arankalle1, Mihai Glont1; Tung Nguyen1, Henning
Hermjakob1
1European Bioinformatics Institute (EMBL-EBI) (UK)
Systems biology models of cell signalling, metabolic and gene regulatory networkshave been shown to divulge mechanistic insight into cellular regulation. One of themajor bottlenecks in building systems models is identification of model parameters.Searching for model parameters from published literature and models is essential,yet laborious task. To address this, we have developed a new resource, BioModelsParameters, that can facilitate easy search and retrieval of parameters values frommodels stored in BioModels (Chelliah et al. 2015).
Two universes one world: Community standards vs. formalstandards in systems biology and systems medicine July 18
10:35amH1Martin Golebiewski1
1Heidelberg Institute for Theoretical Studies (HITS) (Germany)
Given the increasing flood and complexity of data in life sciences, standardizationof these data and their documentation are crucial. This comprises the description ofmethods, biological material and workflows for data processing, analysis, exchangeand integration (e.g. into computational models), as well as the setup, handling andsimulation of models. Hence, standards for formatting and describing data, work-flows and computer models have become important, especially for data integrationacross the biological scales for multiscale approaches.
To this end many grassroots standards for data, models and their metadatahave been defined by the scientific communities and are driven by standardizationinitiatives such as COMBINE and others. For providing the potential users with anoverview and comparable information about such standards we develop informationresources, such as the NormSys registry for modelling standards (http://normsys.h-its.org).
For facilitating the integration of data and models, standards have to be harmo-nized to be interoperable and allow interfacing between the datasets. To supportthis, we drive and lead the definition of novel standards of the International Organi-zation for Standardization (ISO) in the technical ISO committee for biotechnologystandards (ISO/TC 276) in order to define a framework and guideline for commu-nity standards and their application. With our activities we aim at enhancing theinteroperability of community standards for life science data and models and there-fore facilitating complex and multiscale data integration and model building withheterogenous data gathered across the domains.
41
Friday
Model curation and annotationJuly 1910:30am
H1 Anand Rampadarath1
1Auckland Bioengineering Institute (New Zealand)
In an attempt to curtail the ongoing problem of irreproducibility of published work,the Center for Reproducible Biomedical Modeling has introduced an annotation andcuration service to help journal authors, reviewers, and editors publish reproducible,reusable models. Manuscripts received from partner journals will be curated to makesure that any author supplied code will faithfully reproduce the results presentedin the manuscript. In this talk I will give a quick description of this curation andannotation process as well as give an update on the service.
Identifiers.org Compact Identifiers for robust data citationJuly 1910:40am
H1 Henning Hermjakob1, Sarala Wimalaratne1, Manuel Bernal Llinares1, JavierFerrer1, Nick Juty2
1European Bioinformatics Institute (EMBL-EBI) (UK), 2University of Manchester (UK)
Compact identifiers have been informally and widely used for referencing life sciencedata for many years, though the practice has been largely been through ad hocimplementations, serving specific use cases. We describe our implementation, whichhas already begun to be adopted by publishers.
Compact Identifiers consist of an (Identifiers.org) assigned unique prefix in com-bination with a locally (database) assigned accession number (prefix:accession).Compact Identifiers are resolved to database records using information that is storedin an underlying Registry, which contains high quality, manually curated informationon over 700 data collections. This information includes the assigned unique prefix,a description of the data collection, identifier pattern, and a list of hosting resourcesor resolving locations. When a Compact Identifier is presented to the Identifiers.orgResolver, it is redirected to a resource provider, taking into consideration informationsuch as the uptime and reliability of all available hosting resources. For example,pdb:2gc4, GO:0006915, doi:10.1101/101279, orcid:0000-0002-5355-2576 etc.
In addition, a formal agreement with N2T resolver, based in California DigitalLibrary has been struck to provide backup resolution services. Users can there-fore resolve Compact Identifiers using Identifiers.org (https://identifiers.org) or N2T(https://n2t.net/) resolvers. This implementation of Compact Identifiers has beenadopted by Nature Scientific Data for data citations when linking to biomedicaldatasets with accession numbers [2].
[1] Sarala M. Wimalaratne et al. Uniform resolution of compact identifiers forbiomedical data. Sci. Data 5:180029 doi:10.1038/sdata.2018.29 (2018)
[2] Open Editorial. On the road to robust data citation. Sci. Data 5:180095doi:10.1038/sdata.2018.95 (2018)
42
Principles for declarative multicellular modelling July 1912:10pmH1Jorn Starruß1, Walter de Back1, Martin Golebiewski2, Lutz Brusch1
1TU Dresden (Germany),2Heidelberg Institute for Theoretical Studies (HITS) (Germany)
New insights into multicellular processes in tissues and organs, like tissue regen-eration, can be gained using spatially resolved modelling and simulation. Corre-spondingly, many international Systems Medicine projects are developing spatiallyresolved multicellular models and new simulation software. Exchange, reproducibil-ity and archiving of spatially resolved multicellular models among different projectswith different software tools would mean a great leap forward for the community.However, that would require an appropriate and fully declarative model definitionlanguage for this class of models. We present our considerations for the design ofsuch a modelling language based on our declarative modelling experience with Mor-pheus(ML) [1] and highlight central concepts to represent the multicellular complex-ity.
[1] Morpheus(ML) - http://morpheus.gitlab.io
Flapjack: an open-source tool for storing, visualising,analysing and modelling kinetic gene expression data July 19
12:20pmH1Guillermo Yanez1, Isaac Nunez1, Tamara Matute1, Fernan Federici1, Timothy
Rudge1
1Pontificia Universidad Catolica de Chile (Chile)
Engineering design cycles based on accurate parameter estimation from experimen-tal data are key for predictable assembly of complex genetic circuits. In particular,as dynamical systems the reliable design of genetic circuits requires analysis of ki-netic gene expression data. This data is often distributed across many institutions,in different file formats, repositories and databases, which makes it difficult to col-late and reduces the power of analysis. Thus there is a need for data repositoriesthat can store kinetic gene expression data, link this data to circuit designs, andallow analysis that combines multiple studies and experiments to reliably estimateparameters. Here we present Flapjack, a web-based open-source tool for storing,visualising, analysing and modelling kinetic gene expression data. Flapjack enablesusers to flexibly query experimental data based on metadata, for example extract-ing all measurements related to a particular DNA sequence, all growth curves for agiven strain, etc. These queries return all relevant data irrespective of the particularstudy or experiment, or may be restricted to specific experiments of interest. Usingthe web app users may then visualize the experimental time series using interac-tive plots (time courses, kymographs, heatmaps), apply analyses such as calculationof expression rates, growth rates, and parameterise transfer functions or inductioncurves. For custom analysis queried data can be downloaded in CSV, JSON orXML file formats. Currently, Flapjack supports upload of Synergy HTX and BMGLabtech microplate reader data, but can easily be extended to accommodate any
43
data format. We propose a data repository and set of analysis tools that enablesscaleable data analysis, visualization and parameter estimation, collating data be-tween institutions, studies, experiments, and individual researchers. Flapjack willthus significantly enhance data sharing, management and analysis for synthetic andsystems biology.
44
Workshops and Breakouts
Tuesday
Sharing experiences in building a standards community July 1614:00pmH1Dagmar Waltemath1, Mike Hucka2
1University Medicine Greifswald (Germany),2Caltech, Pasadena (USA)
COMBINE is now 10 years old and consists of 8 core standards, plus an additional5 associated standards. Some of these standards have a much longer history. Com-mon to all standards is their bottom-up, community-based approach to developing,evaluating and refining the information to be shared about a piece of research incomputational biology modeling and related fields. Our experience when presentingCOMBINE at workshops, conferences and meetings is that there is substantial in-terest in how communities can be built in such a sustainable and open manner. Inthis breakout session, we invite long-term members of the community to get togetherand collect with us hints & tips for community building in a scientific, data-drivenfield of research. We intend to provide the tips to the public afterwards, hoping thatothers can benefit from our experiences. As an additional topic of discussion, we willdiscuss how we can improve our communication, not only about community-buildingprocesses across different scientific fields, but also about the benefits of COMBINEitself in research fields relating to computational modeling in biology.
SED-ML Script: a Proposal July 1614:00pmH2Lucian Smith1, Herbert Sauro1
1University of Washington (USA)
SED-ML was first introduced 10 years ago. As the next levels and versions are beingdeveloped, perhaps it is time to consider a procedural format, instead of the existingdeclarative format. SED-ML Script is a developing project by Herbert Sauro andLucian Smith at the University of Washington that attempts to take what we’velearned in the past ten years and turn it into a script form. Discussion encouraged.
45
COMBINE as Legal EntityJuly 1616:00pm
H1 Martin Golebiewski1, Herbert Sauro2
1Heidelberg Institute for Theoretical Studies (HITS) (Germany), 2University ofWashington, Seattle, WA (USA)
Options for the COMBINE community to become a legal entity will be discuss. Theaim of the session is to develop a roadmap for a sustainable future for the COMBINEnetwork. An expected outcome of the meeting is to decide on concrete actions tobe taken in order to establish a legal COMBINE entity. Such a legal entity canappear as partner in research proposals and receive corresponding funding for thefurther development, implementation and promotion of modelling standards in thelife sciences. A COMBINE entity can thus provide a formal home for the COMBINEcore standards and comparable standardization activities.
46
Wednesday
SBGN workshop July 1714:00pmH1Falk Schreiber1, Michael Blinov2
1University of Konstanz (Germany), 2UConn School of Medicine (USA)
The following topics will be discussed at the workshop:
• Tools and databases supporting SBGN
• Features that SBGN need
• Specifications
• Outreach and funding
See the workshop webpage for details: https://sbgn.github.io/sbgn12
MIRIAM 2 & the OMEX Metadata Specification July 1714:00pmH2Dagmar Waltemath1, David Nickerson2
1University Medicine Greifswald (Germany), 2University of Auckland (New Zealand)
The Minimum Information Requested In the Annotation of Biochemical Models(MIRIAM) is a widely accepted and referenced Minimum Information Guideline forannotating computational models. MIRIAM was released in 2005 and has been ad-hered to by major curation services and modeling teams. However, with the creationof the OMEX archive, models are no longer necessarily at the center of a sharedresearch project. Instead, collections of files belonging to a virtual experiment offera way to package a model with all necessary information for reproducing simulationexperiments. Recently, COMBINE community members published recommenda-tions for harmonizing semantic annotations stored in OMEX archives, offering aconsensus approach for capturing the meaning of elements within the various typesof resources shared in these archives. We propose to run a breakout session at COM-BINE to discuss updating the MIRIAM guidelines in light of the recent semanticannotation recommendations as well as how to implement the revised guidelinesin annotation software packages. We invite representatives from all modeling stan-dards, curators from model repositories, modelers, and other interested parties for aninitial brainstorming session and a potential launch of a working group to coordinatethe revision of the original MIRIAM guidelines.
47
SBOL breakoutJuly 1714:00pm
H3 Ernst Oberortner1, Chris Myers2
1DOE Joint Genome Institute (JGI), Berkeley Labs (USA), 2University of Utah (USA)
Model eXchange consortium: Inaugural meetingJuly 1916:00pm
H2 Rahuman Sheriff1, Henning Hermjakob1
1European Bioinformatics Institute (EMBL-EBI) (UK)
Model repositories play an essential role in making models easily sharable andaccessible. Searching for a model requires users to visit multiple repositories andlook for a model of their interest. A common platform that will allow modellers tosearch for models across all existing repositories is the need of the hour. This re-quires model repositories to share the model metadata in a common platform wheresearches can be performed by users. Following the search, users can be redirectedto the appropriate repository to download the models of their interest. Henning’spresentation at HARMONY 2019 to form a consortium of modelling repositoriesto collaboratively develop a common search platform gained interest from many.Model exchange consortium will also facilitate development of common and sharedcuration standards and pipelines as well as opportunity for repositories to supporteach other. To move forward, we agreed to invite all interested repositories andorganise an inaugural meeting for the Model eXchange consortium in a breakoutsession at COMBINE 2019.
48
Thursday
EU-STANDS4PM workshop July 1814:00pmH1Marc Kirschner1, Martin Golebiewski2
1Forschungszentrum Julich (Germany), 2Heidelberg Institute for Theoretical Studies(HITS) (Germany)
One of the major goals of EU-STANDS4PM is to assess and evaluate national stan-dardization strategies for interoperable health data integration as well as data-drivenin silico modelling approaches for personalized medicine with the aim to bundle Eu-ropean standardization efforts. The project will produce an in depth EU-wide map-ping of relevant European initiatives with regard to data sources (WP1) as well as insilico models (WP2). This process is the foundation to assemble specific recommen-dation and guidelines (including EU standard documents) for data harmonizationand integration strategies as well as data-driven in silico approaches to interprethuman disease/health data.
The current EU-STANDS4PM workshop at the annual COMBINE meeting isembedded in the context of work package 1 ”Data sources and standards for pre-dictions in personalized medicine” with the objective to initiate and consolidatea pan-European standardization framework for in silico methodologies applied inpersonalized medicine. A central aim of WP1 is to drive harmonization and inter-operability of domain-specific standards (scientific bottom-up/community standardsand standards set by national and international standardization bodies) in person-alized medicine in order to facilitate data integration to enable predictive in silicomodeling on a broader European level.
Through the workshop EU-STANDS4PM will consult the COMBINE communityto put a focus on (i) analyzing interoperability and scalability of data and metadatastandards relevant in COMBINE and (ii) reflect on possibilities for cross-domainand cross-technology data integration to facilitate in silico modelling approaches inpersonalized medicine.
To discuss these topics interactive parallel discussions in smaller groups (”worldcafes”) will collect and debate on several key points:
• Data and model standards - Which of them are relevant for personalizedmedicine
• Reproducibility - Standards as drivers
• Integration of clinical and research data - Approaches, problems and standard-ization gaps
• Pitfalls in developing and harmonizing standards
• Transformation/Transition from community standards to formal standards
• Using patient-derived data for personalized medicine: Legal and ethical as-pects
49
Something old, something new: constraint-based modellingand SBMLJuly 18
14:00pmH2 Brett G. Olivier1, Frank T. Bergmann2
1AIMMS VU University Amsterdam (Netherlands),2 BioQuant/COS, HeidelbergUniversity (Germany)
The SBML Level 3 Flux Balance Constraint package (FBC) is now swiftly moving to-wards its third https://www.overleaf.com/project/5d15d2564d30874b7e91ed70 ver-sion. The Version 3 iteration further extends the types of constraint-based modelsthat can be encoded in SBML. Furthermore, new elements are introduced that en-able the encoding and annotation of genome-scale, metabolic reconstructions. Asthis is the FBC package’s 10th birthday we also reflect on how the developmentof the FBC package has mediated a productive collaboration between two diversecommunities. Leveraging the strengths of this collaboration has led to FBC beingone SBML’s most actively developed packages.
SBML L3 Packages - An Introduction to SBML packagesJuly 1816:00pm
H2 Sarah Keating1
1University College London (UK)
SBML is the community-developed format for enabling the exchange and reuse ofcompuational models in biology. SBML Level 3, has a modular structure, with acore suited to representing reaction-based models, and packages that extend thecore with features suited for a variety of model types. This allows the encodingof constraint-based models, reaction-diffusion models, logical network models, andrule-based models. Many L3 packages have been finalised, a couple are approachingthe final stage of the process and there are still some packages where more work isrequired.
The development process requires a written specification and two implementa-tions that can demonstrate exchange of models and reproducibility of results, ifappropriate. The SBML Editors then review the document and implementations inorder to ratifiy the package.
This talk will give a brief overview of the development process and the capabilitiesof all SBML L3 packages, both those that have been finalised and those still indevelopment. The floor will then be open to continue the discussion of packagesapproaching finalisation: Spatial Processes and Distributions; discussion of packagesrequiring more input: Arrays; and discussion of packages that are not currentlyprogressing: Dynamic Processes and Math. The aim of the session is to move thingsforward or identify and propose solutions to any impediments.
Participants are also welcome to suggest improvements and/or additional pack-ages that are necessary for their own modelling.
50
Friday
FAIRDOM PALs and User Workshop July 1914:00pmH1Olga Krebs1
1Heidelberg Institute for Theoretical Studies (HITS) (Germany)
FAIRDOM is a research infrastructure offering data management support. TheSEEK software is designed as registry and storage place for data, models, biolog-ical samples, processes, publications and presentations, and at the same time asyellow pages for projects, people and events. SEEK is implemented as central datamanagement platform FAIRDOMhub (https://www.fairdomhub.org).
In this meeting, FAIRDOM team will present project’s current status and futureplans, high-profile users of FAIRDOM software will share their data managementpractices. Finally, we will hold informal discussions and breakouts where the au-dience, and their needs will define what we discuss. Detailed agenda https://fair-dom.org/events/fairdom-pals-users-meeting-2019/
You will have direct contact with the FAIRDOM Community and Tech Team,who can give you personalised advice on your data and model management, andusing FAIRDOM.
The meeting welcomes everybody whether you are a PhD student or a professor,a current FARIDOM user or just curious about data and model management.
Model eXchange consortium: Inaugural meeting July 1916:00pmH2Rahuman Sheriff1, Henning Hermjakob1
1European Bioinformatics Institute (EMBL-EBI) (UK)
Model repositories play an essential role in making models easily sharable andaccessible. Searching for a model requires users to visit multiple repositories andlook for a model of their interest. A common platform that will allow modellers tosearch for models across all existing repositories is the need of the hour. This re-quires model repositories to share the model metadata in a common platform wheresearches can be performed by users. Following the search, users can be redirectedto the appropriate repository to download the models of their interest. Henning’spresentation at HARMONY 2019 to form a consortium of modelling repositoriesto collaboratively develop a common search platform gained interest from many.Model exchange consortium will also facilitate development of common and sharedcuration standards and pipelines as well as opportunity for repositories to supporteach other. To move forward, we agreed to invite all interested repositories andorganise an inaugural meeting for the Model eXchange consortium in a breakoutsession at COMBINE 2019.
51
52
Lightning Talks
Systems Biology Graphical Notation (SBGN) 15 July14:30amH1Michael Blinov1
1UConn School of Medicine, Farmington (USA)
The Systems Biology Graphical Notation (SBGN), is a set standard graphicallanguages to describe visually biological knowledge. It is currently made up ofthree languages describing Process Descriptions (PD), Entity Relationships (ER)and Activity Flows (AF).
In this lightning talk, I will briefly summarize the latest edition of SBGN.
SBML Level 3 Version 2 and SBML Level 3 packages 15 July14:30amH1Michael Hucka1
1Caltech, Pasadena (USA)
The Systems Biology Markup Language (SBML) project is an effort to developa machine-readable exchange format for computational models in biology. By sup-porting SBML as an input and output format, different software tools can operateon the same representation of a model, removing chances for errors in translationand assuring a common starting point for analyses and simulations. Today, SBMLis the de facto standard in this area, and it is continually being evolved to sup-port more types of models and modeling needs by a community of developers andsupporters worldwide.
While SBML was initially developed to exchange non-spatial compartmentalmodels of biochemical reaction networks primarily formulated in terms of chemicalkinetics, it was always understood that there existed more types of models. Overtime, SBML was expanded to support a broader range of model types, modelingparadigms, and research areas. SBML Level 3 introduced an extensible modulararchitecture consisting of a central set of fixed features (named SBML Level 3 Core),and a scheme for adding ”packages” that can augment the Core by extending existingelements, adding new elements, and adjusting the meaning or scope of elements.This permit layering the core of SBML with new features suited to more types ofmodels, together with a way for individual models to identify which sets of extensionsthey need for proper interpretation. Twelve SBML Level 3 packages have beenproposed to date.
In this lightning talk, I will briefly summarize the latest edition of SBML Level3 and the current SBML Level 3 packages.
53
WebProv: A web-based tool to access, store, and displayprovenance information of simulation modelsJuly 15
14:30pmH1 Kai Budde1, Jacob Smith2, Andreas Ruscheinski1, Adelinde M. Uhrmacher1
1University of Rostock (Germany), 2University of New Brunswick (Canada)
Provenance provides ’information about entities, activities, and people involvedin producing a piece of data or thing, which can be used to form assessments aboutits quality, reliability, or trustworthiness’ [1]. For simulation models, provenanceincludes, for example, information about simulation experiments that have beenexecuted, data that has been used as input for calibration or validation of the simu-lation model, or other simulation models the simulation model has been based upon[2]. We will present and discuss a first working version of a web-based provenancetool to access, store, and display provenance information of simulation models. Thefront end includes a query tool that allows the user to search for different keys (e.g.,name of cell line) and model dependencies (e.g., model extensions) and displays re-sults graphically and as text. As an example, we relate different Wnt models to oneanother by capturing provenance information using the PROV data model standardand show detailed graphs for some of the chosen Wnt models.
[1] Groth, P., Moreau, L.: Prov-overview. an overview of the prov family ofdocuments (2013), https://www.w3.org/TR/prov-overview/.
[2] Andreas Ruscheinski, Dragana Gjorgevikj, Marcus Dombrowsky, Kai Budde,and Adelinde. M. Uhrmacher. Towards a PROV ontology for simulation models.In International Provenance and Annotation Workshop, pages 192-195. Springer,2018.
MIRIAM 2 & the OMEX Metadata SpecificationJuly 1514:30pm
H1 Dagmar Waltemath1, David Nickerson2
1University Medicine Greifswald (Germany), 2University of Auckland (New Zealand)
see page 47
54
A Virtual Cohort of Heart Failure Patients Four-chamberHeart Meshes for Cardiac Electro-mechanics Simulations July 15
14:30pmH1Marina Strocchi1, Christoph Augustin2, Matthias Gsell2, Orod Razeghi1, Anton
Prassl1, Edward Vigmond3, Jonathan Behar4, Justin Gould4, Sidhu Baldeep4, AldoRinaldi4, Martin Bishop1, Gernot Plank2, Steven Niederer1
1King’s College London (UK), 2Medical University of Graz (Austria), 3University ofBordeaux and LIRYC Electrophysiology and Heart Modeling Institute (France), 4King’s
College London and Guy’s and St Thomas’ NHS Foundation Trust (UK)
Computational models of the heart are increasingly being used in the develop-ment of devices, patient diagnosis and therapy guidance. While software techniqueshave been developed for simulating single hearts, there remain significant challengesin simulating cohorts of virtual hearts from multiple patients. To facilitate the devel-opment of new simulation and model analysis techniques by groups without directaccess to medical data, image analysis techniques and meshing tools we have createda virtual cohort of fifteen hearts. Our cohort was built from heart failure patients,age 6815 years and 14 males. We segmented four-chamber heart geometries fromend-diastolic CT images. We generated tetrahedral meshes with an average edgelength of 1.10.2mm. We added ventricular fibres with a rule-based method withan orientation of -60 and 80 at the epicardium and endocardium, respectively. Weran benchmark electromechanics simulations to test the numerical stability of themeshes. We simulated ventricular electrical activation with a reaction-eikonal model.Passive mechanics of ventricular tissue was represented with a transversely-isotropicGuccione law. All the other tissues were modelled as isotropic. Ventricular preloadand afterload were modelled as constant atrial pressure and three-element Wind-kessel models, respectively. Omni-directional springs were applied at the croppedleft pulmonary veins and at the superior vena cava. Normal springs were appliedat the ventricular epicardium to represent the effect of the pericardium. The sameparameters were used for the fifteen meshes. The electrophysiology simulations re-sulted in a QRS duration of 13413 ms. The mechanic simulations resulted in a LVand RV ejection fraction of 383 and 283 % and peak LV and RV systolic pressure of983 and 241 mmHg. Our results prove that our four-chamber meshes are suitablefor electromechanics simulations. The virtual cohort will be made publicly availablein a future publication.
WebLab update July 1514:30pmH1Sarah Keating1
1University College London (UK)
WebLab is an online tool that allows users to upload mathematical models andtext-based descriptions of experiments (protocols) and run/compare results of usingdifferent combinations of the two. We have recently added the ability to uploadexperimental data and perform fitting of mathematical models to these data in areproducible fashion.
55
NESTML: A domain-specific language for biologicallyrealistic neuron and synapse modelsJuly 15
14:30pmH1 Charl Linssen 1, Jochen M. Eppler1, Abigail Morrison2
1Forschungszentrum Julich (Germany), 2Institute of Cognitive Neuroscience,Ruhr-University Bochum (Germany)
NESTML [1, 2] was developed to address the maintainability issues that follow froman increasing number of models, model variants, and an increased model complex-ity in computational neuroscience. Our aim is to ease the modelling process forneuroscientists both with and without prior training in computer science. This isachieved without compromising on performance by the use of automatic source-codegeneration. While originally developed in the context of the NEST Simulator [31],the language itself as well as the associated toolchain are lightweight, modular andextensible, by virtue of using a parser generator and internal abstract syntax tree(AST) representation. A typical workflow begins by identifying a model of interest,for example, the dynamical behavior of a single neuron, or the plasticity rules con-cerning a synapse. This model might come in mathematical or textual form, but issubsequently formalised by the neuroscientist following the NESTML syntax. Modelentry is facilitated by a compact syntax and language features such as static and dy-namic typing, integrated support for physical units, and the ability to directly enterdynamical equations in differential form. Next, invoking the toolchain will generateoptimised code for a specified target platform. That code is dynamically loaded orcompiled as part of the simulation framework, after which it can be instantiatedwithin a simulation script, written e.g. using the PyNEST API [4], before startingthe simulation. NESTML is open sourced under the terms of the GNU General Pub-lic License v2.0 and is publicly available at https://github.com/nest/nestml. Exten-sive documentation and automated testing are in place, both for the language itselfas well as the associated processing toolchain. Active user support is provided viathe GitHub issue tracker and the NEST user mailing list. Acknowledgements: Thisproject has received funding from the Helmholtz Association through the HelmholtzPortfolio Theme ”Supercomputing and Modeling for the Human Brain” and theEuropean Union’s Horizon 2020 research and innovation programme under grantagreement No 720270 (HBP SGA1) and No 785907 (HBP SGA2).
1. D. Plotnikov et al. (2016) Modellierung March 2-4 2016, Karlsruhe, Germany.
2. K. Perun et al. (2018). Version 2.4, Zenodo.
3. M.-O. Gewaltig and M. Diesmann (2007) Scholarpedia 2(4), 1430.
4. Y.V. Zaytsev and A. Morrison (2014) Front. Neuroinform. 8:23
56
BioModels 2019 July 1514:30pmH1Mihai Glont1, Henning Hermjakob1, Rahuman Sheriff1, Krishna Tiwari1,2, Tung
Nguyen1, Manh Tu Vu1
1European Bioinformatics Institute (EMBL-EBI) (UK), 2Babraham Institute (UK)
BioModels is a repository of mathematical models describing biological processes.In this talk I will summarise recent developments.
Minibatch optimization: a method for training mechanisticmodels using large data sets July 15
14:30pmH1Paul Stapor1, Leonard Schmiester, Jan Hasenauer, Daniel Weindl
1Institute of Computational Biology, Helmholtz Zentrum Munchen (Germany)
In recent years, a number of rich public data bases, such as the Cancer Cell LineEncyclopedia (CCLE), have emerged. They constitute a powerful tool for compu-tational studies of complex diseases such as cancer and make it possible to developcomprehensive mechanistic models, e.g., based on ordinary differential equations(ODEs). Such models are necessary for a detailed understanding of the underly-ing cellular processes. However, when dealing with ODE-models and big data sets,parameter estimation becomes the limiting factor in model development, as compu-tation time increases linearly with the number of experimental data sets used formodel training. This prohibits the use of ODE-models to either small models orlimits amount of data to train the model on. Mini batch optimization methods canbe employed to circumvent the linear scaling of computation time with the numberof the data sets. Although being widely used for model training of deep neural nets,these methods have never been applied for the training of ODE-models, to the bestof the authors knowledge. One reason for this may be that this transfer to anotherscientific field is not straight forward. Typical optimization problems have differ-ent properties in both fields, e.g., in ODE-modeling, the underlying ODEs can benumerically not solvable for certain regions of the parameter space. This makes itnecessary to adapt mini batch optimization methods for the application to ODE-model training. We show that it is possible to transfer some of the most commonmini batch optimization algorithms to the field of ODE-model training. Using ade-quate adaptations, we can reduce computation and wall time for model training oflarge-scale ODE-models by up to an order of magnitude, while keeping good con-vergence properties. We moreover present additional approaches, which may helpto further improve the performance of the presented methods.
57
Modelling pancreas - Computational modelling and analysisof altered metabolism in the pancreas and its relationship
to diseaseJuly 1517:00pm
H1 Deepa Maheshvare1 Soumyendu Raha1, Matthias Konig2, Debnath Pal1
1Indian Institute of Science (India), 2Humboldt-University Berlin (Germany)
An important aspect of many non-infectious diseases are alterations in metabolism.These alterations are a key feature of metabolic disorders such as Diabetes, Obe-sity, Cardiovascular Disorders, Fatty Liver Disease and arguably, Cancer. Here, wepresent a model of the islets of Langerhans to study alterations of Human glucosemetabolism due to altered pancreatic function (e.g., impaired glucose tolerance anddiabetes). Diabetes is characterized by the occurrence of hyperglycemic events as aconsequence of altered metabolic processes that control the secretion and uptake ofinsulin by the pancreatic islets and hepatocytes, respectively. Our metabolic modelof the pancreatic beta-cell includes the uptake of nutrients by the cell from the blood,the metabolic processes that trigger the release of hormones like insulin into theblood, and a model of insulin secretion. Our kinetic framework deploys multi-omicsdata derived from in vitro and in vivo experiments to incorporate human-specificgene-protein-reaction data to assess the transient change in the concentration ofbiochemical species. Additionally, we apply transcriptomics data to scale the re-action velocities for assessing how the dysregulation of enzymes that regulate theflux through metabolic pathways such as carbohydrates and amino acids, which actas insulin secretagogues, control the exocytosis of insulin granules from the isletsof pancreas. Our model will allow us to eventually explore the alterations in Hu-man glucose metabolism that arise due to perturbations in metabolic flux in thepancreatic islets in a disease state versus a healthy state, in a systematic manner.
PK-DB 1.0: a pharmacokinetics database for individualizedand stratified modelingJuly 15
17:00pmH1 Jan Grzegorzewski1, Matthias Konig1
1Humboldt University Berlin (Germany)
We present PK-DB, a database for the representation of pharmacokinetics data fromclinical trials and pre-clinical research. Data is either curated from the literatureor from available raw data. The main focus of PK-DB is to provide high-qualitypharmacokinetics data in combination with required meta-information for compu-tational modeling and data integration, i.e., (i) characteristics of studied patientcollectives and individuals; (ii) applied interventions (dosing, route, ); and (iii) mea-sured pharmacokinetics information and time courses. Important features are therepresentation of experimental errors and variation, the representation and normal-ization of units, annotation of information to biological ontologies, and calculationof pharmacokinetics information like apparent clearance, half-life, or area under thecurve (AUC) from time course data. We demonstrate the value of PK-DB by (i)a stratified meta-analysis of pharmacokinetics studies for caffeine curated from the
58
literature, thereby integrating pharmacokinetics information from a wide range ofsources; (ii) providing examples for the computational modeling of dynamical liverfunction tests based on PK-DB data.
Standardised meta-data in Hospital Information Systemsunlock clinical data for research purposes A real-world
example July 1517:00pmH1Robert Gott1, Kai Fitzer1, Torsten Leddig1, Dagmar Waltemath1, Thomas
Bahls1
1University Medicine Greifswald (Germany)
Only few commercial Hospital Information Systems (HIS, German: KAS) providemeans to encode and use biomedical meta-data thus allowing for Good Epidemi-ological Practice (GEP) [1]. Moreover, HIS typically encode meta-data as part ofthe core system, making extraction, reuse, and linkage of meta-data with other datasources a difficult task. A separated, standards-aware or even standards-compliantmeta-data repository is more desirable: the meta-data is easily accessible and canbe queried by clinical as well as research data providers or consumers. Consistentlinkage of data items guarantees fast access to the original data sets as well as tofurther resources of the HIS environment. Such a knowledge base has the potentialto better connect the worlds of systems medicine and medical informatics, and opensup new opportunities for collaborative research in the biomedical domain. As onepositive example, the KAS+ project, initiated at the University Medicine Greifswald(UMG), provides an integrated IT infrastructure for hospital (KAS) and researchunits (the Plus) [2]. It connects patient care and research, making patient care dataavailable for research purposes, and supporting research projects such as clinicalstudies. A central metadata repository (MDR) stores meta-data of the multitudeof heterogeneous clinical data sources, for example using ISO/IEC 11179. Manu-facturers of all components of the IT infrastructure supply their own metadata forthe MDR in a compatible digital format. Hence, the KAS+ data can be searched,and the extracted data can be linked to external resources (e.g., biobanks, reposi-tories of computational simulation models, or analysis software). On our poster, wedemonstrate how meta-data in the MDR is structured, and how it is used to linkheterogeneous data sets when addressing complex research questions.
[1] Hoffmann, W., Latza, U., Baumeister, S.E., Brnger, M., Buttmann-Schweiger,N., Hardt, J., Hoffmann, V., Karch, A., Richter, A., Schmidt, C.O. and Schmidt-mann, I., 2019. Guidelines and recommendations for ensuring Good Epidemiologi-cal Practice (GEP): a guideline developed by the German Society for Epidemiology.European journal of epidemiology, pp.1-17.
[2] Hoffmann W, van den Berg N, Fitzer K, Leddig T, Jackisch M, Bahls T;Enabling Healthcare Data for Health Services Research Recent Developments inApplied Medical Informatics in Germany.
59
Development of python visualization package DNAplotlibJuly 1517:00pm
H1 Sunwoo Kang1
1Stanford University (USA)
DNAplotlib is a quick, automated visualization package to support visualizationof genetic circuits (mainly transcription & translational level). The current modelfocuses on visualizing SBOL files, a file type that is widely used in the field of syn-thetic biology. In the time of python becoming the standard tool for data analysis,DNAplotlib offers a facilitation of communication between researchers and develop-ers about their genetic modules used in their lab.
Cy3SBML: a Cytoscape app for SBMLJuly 1517:00pm
H1 Matthias Konig1, Nicolas Rodriguez2, Andreas Drager3,4,5
1Humboldt-Universitat zu Berlin (Germany), 2Babraham Institute, Cambridge (UK),3University of Tubingen and German Center for Infection Research (DZIF) (Germany)
We present Cy3SBML, a Cytoscape app for the community format SBML. Key fea-tures include (i) SBML import of all levels and versions; (ii) support for SBML level 3packages, specifically for the SBML extensions for Model Layouts (layout), Qualita-tive Models (qual), Groups (groups), Hierarchical Model Compositions (comp) andFlux Balance Constraints (fbc), (iii) navigation and manipulation of network lay-outs based on SBML structure, (iii) access to RDF model annotations in MIRIAMformat as well as to SBO-based annotations, (iv) REST API for automatization,and (v) validation of SBML. Cy3SBML provides a variety of importers for SBMLmodels with direct access to BioModels and the BiGG database via web services.Cy3SBML is available from the Cytoscape app store and GitHub.
60
Posters
Review of software tools and data sources supporting theSystems Biology Graphical Notation (SBGN) standard Poster 1
PostersessionsFoyer
Michael Blinov1, Adrien Rougny2, Andreas Drager3, Vasundra Toure4, UgurDogrusoz5, Alexander Mazein6
1University of Connecticut School of Medicine (USA),2Biotechnology Research Institutefor Drug Discovery National Institute of Advanced Industrial Science and Technologyand Computational Bio Big-Data Open Innovative Laboratory (Japan), 3University of
Tubingen (Germany), 4Norwegian University of Science and Technology (Norway),5Bilkent University (Turkey), 6University of Luxembourg (Luxembourg)
Visualisation is a critical part of understanding and conveying biological informa-tion. The Systems Biology Graphical Notation (SBGN) is a standard developed toprovide a notation, as well as requirements and guidelines, for the visual represen-tation of biomolecular networks. As for the XML-based SBGN Markup Language(SBGN-ML), it was designed to facilitate storage and exchange of SBGN diagrams.Over time, SBGN was adopted by multiple software tools and data sources, allowingbiological researchers to create and edit SBGN-compliant diagrams, as well as visu-alize data originally represented in other systems biology formats (such as SBML,BioPAX, KEGG-ML). We will review software tools and data sources supportingSBGN and SBGN-ML.ions made how we could tackle the still existing problems.
WebProv: A web-based tool to access, store, and displayprovenance information of simulation models Poster 2
PostersessionsFoyer
Kai Budde1, Jacob Smith2, Andreas Ruscheinski1, Adelinde M. Uhrmacher1
1University of Rostock (Germany), 2University of New Brunswick (Canada)
see page 53
61
Comprehensive mathematical modeling of sphingolipidmetabolism by integration of lipidomics and proteomics
dataPoster 3Poster
sessionsFoyer
Canan Has1, Thorsten Hornemann2, Robert Ahrends1
1Leibniz-Institut fur Analytische Wissenschaften - ISAS - e.V (Germany), 2UniversityHospital Zurich (Switzerland)
Sphingolipids play an important role in cell membrane composition and lipid raftformation. In addition to that, they exert a vital role in regulating multiple cellularfunctions including signaling, apoptosis, inflammation. Sphingolipids show antago-nistic activities. While ceramide and sphingosine are proapoptotic to benefit growtharrest and apoptosis, ceramide-1-phosphate and sphingosine-1-phosphate are knownto promote cell proliferation, transformation and inflammation. The metabolism ofsphingolipids, including the all enzymes and lipids involved, has been extensivelystudied. This metabolic pathway delineates an integrated system linking multiplesynthesis and catabolism pathways by centering ceramide. The three main path-ways that compose sphingolipid metabolism are de novo synthesis, the hydrolysisof sphingomyelin and recycling of gangliosides also called salvage pathway. Thealteration in sphingolipid metabolism results in pathogenicity of several diseasesincluding neurodegenerative diseases and metabolic diseases. Hence, integrationof quantitative lipid and protein data along with biochemical reaction kinetics tomodel sphingolipid metabolism would be crucial to shed light on sphingolipid re-lated disease mechanisms. Here, we propose a comprehensive and extended modelof sphingolipid metabolism in mouse. Contrary to the previous studies, we incor-porated different long-chain based fatty acid preferences in de novo synthesis ofceramide, considered substrate affinities of enzymes and cross-talk. The rate con-stants were estimated through parameter estimation via COPASI and the resultingmodel was fit experimental data for all species. The model validation was performedon lipidomics and proteomics data of RAW264.7 macrophage cell activated by anendotoxin, Kdo2-Lipid A available in the LIPID MAPS consortium. The model wepresent here will further contribute to our understanding of sphingolipid metabolismand will support to unravel the behavior of sphingolipids in disease models such asAlzheimer and insulin-resistance.
The Systems Biology Graphical Notation: a standardisedrepresentation of biological mapsPoster 4
Postersessions
Foyer
Andreas Drager1, Vasundra Toure2, Alexander Mazein3, Adrien Rougny4, UgurDogrusoz5, Michael Blinov6, Augustin Luna7
1University of Tubingen (Germany), 2Norwegian University of Science and Technology(Norway), 3Universite de Lyon (France),4National Institute of Advanced IndustrialScience and Technology (Japan), 5Bilkent University (Turkey), 6UConn School of
Medicine (USA), 7Harvard Medical School (USA)
Background: Visualization of biological processes plays an essential role in life sci-ence research. Over time, diverse forms of diagrammatic representations, akin to
62
circuit diagrams, have evolved without well-defined semantics potentially leading toambiguous network interpretations and difficult programmatic processing.
Results: The Systems Biology Graphical Notation (SBGN) is a standard de-veloped to reduce ambiguity in the visual representation of biomolecular networks.It provides specific sets of well-defined symbols for various types of biological con-cepts. SBGN comprises three complementary languages: Process Description (PD),Entity Relationship (ER), and Activity Flow (AF). SBGN PD is based on reactionsand is well-suited for detailed sequential biochemical mechanisms, for instance, torepresent metabolic pathways. SBGN AF shows cascades of influences between theactivities carried by biomolecular entities (e.g., stimulation, inhibition) and is par-ticularly useful when the precise molecular mechanisms are unknown or do not needto be shown, for instance, to represent signalling pathways and regulatory networks.SBGN ER represents independent interactions between features of biological entities,which avoids combinatorial explosions of represented biological states and interac-tions. The XML-based SBGN Markup Language (SBGN-ML) facilitates convenientstorage and exchange of SBGN maps, supported by the library libSBGN.
Discussion: The SBGN project is an ongoing open community-driven effort co-ordinated and maintained by an elected international editorial board. Annual work-shops, GitHub and mailing lists are used as leading discussion platforms. Majorresearch projects, such as the Virtual Metabolic Human, and pathway databasessuch as Reactome and WikiPathways display their maps following the SBGN guide-lines. Furthermore, a wide range of tools supports SBGN. SBGN regularly offersstudent coding events through the Google Summer of Code program.
Availability: All documents and source code are freely available at http://sbgn.org and https://github.com/sbgn. Contributions are welcome.
Contact: [email protected]: SBGN, circuit diagram, biological network, visualisation, systems
biology
AMICI and PESTO - A strong couple for simulation andparameter estimation in computational and systems biology Poster 5
PostersessionsFoyer
Erika Dudkin1, Paul Stapor2,3, Daniel Weindl2, Benjamin Ballnus2,3, SabineHug2,3, Carolin Loos2,3, Anna Fiedler2,3, Sabrina Krause2,3, Sabrina Hroß2,3,Yannik Schalte2,3, Leonard Schmiester2,3, Dantong Wang2,3, Elba Raimndez
Alvarez2,3, Simon Merkt1, Fabian Frohlich2,3, Jan Hasenauer1,2,3
1University of Bonn (Germany),2Helmholtz Zentrum Munchen (Germany), 3TechnischeUniversitat Munchen (Germany)
Mechanistic models in systems biology are important to gain an understandingof the physical processes in biological systems. Those models are often based onordinary differential equations. Since large scale-models become increasingly impor-tant as more and more data is available, powerful toolboxes are needed. AMICI(Advanced Multilanguage Interface to CVODES and IDAS) is a toolbox employingthe numerical solvers CVODES and IDAS from the C-based SUNDIALS package.It has an easy-to-use Python/MATLAB/C++ interface, while ODE integration is
63
shifted to C to gain speed. Model setup can be done in Python/MATLAB, but alsomodels in the common SBML format can be imported. AMICI can compute firstand second order derivatives of functions depending on the ODE solution. To dothis accurately and rapidly, it can perform forward and adjoint sensitivity analysis infirst and second order. This makes AMICI applicable for large scale problems. AM-ICI is furthermore able to deal with discrete events and perform sensitivity analysisfor them. Events can have particular observables and may depend on the simu-lated time, but also on the system state or model parameters, which is a uniquefeature of AMICI. Parameter estimation is a common problem in systems biology.Often, unknown parameters have to be inferred from measurement data. This leadsto non-convex and possibly multi-modal optimization problems. PESTO (Parame-ter EStimation TOolbox) is a highly customizable MATLAB-based toolbox, whichpresents a wide range of state of the art methods for optimization, such as multi-start local, global or hybrid methods, as well as uncertainty analysis, employingprofile likelihoods computed based on different approaches and numerous differentMarkov-Chain Monte Carlo methods. Confidence or credibility intervals can becomputed to user-provided thresholds and all results from either optimization oruncertainty analysis can be visualized using customizable plotting routines. Cur-rently, pyPESTO, a Python-based toolbox is being developed which will have thesame functionalities as PESTO with further features such as supporting PEtab adata format for specifying parameter estimation problems in systems biology, as wellas parallel computing. AMICI and (py)PESTO are available as open source softwareand have been used in several publications. They can be easily interfaced with eachother, making them a strong couple for fitting mechanistic models in systems andcomputational biology.
Analyzing Genetic Circuits for Hazards and GlitchesPoster 6Poster
sessionsFoyer
Pedro Fontanarrosa1, Hamid Hosseini2; Amin Borujeni2, Yuval Dorfan2, ChrisVoigt2, Chris Myers1
1University of Utah (USA), 2MIT, Boston (USA)
see page 31
64
YoMoSt - Statistics for you local model versions Poster 7PostersessionsFoyer
Tom Gebhardt1, Martin Scharm2, Olaf Wolkenhauer3
1University of Rostock (Germany)
Development is often a non-linear process whether you plan buildings, program toolsor create biochemical models. Most of the time, one ends up with a set of versionsfrom different approaches and improvements. Especially modellers of biochemicalnetworks may tend to keep older versions of their models due to the pursuance ofreproducibility. Thus, they can end up with an individual file structure, which growsin its complexity during the model development. This complicates the tracking ofdifferences and the evaluation of results. While the use of repositories improved themanagement of model versions, the investigation of currently unpublished models isstill tedious manual work. With YoMoSt, we present a pipeline of tools to supportstudying of locally stored models. Based on the BiVeS algorithm and as an adaptionof the MoSt pipeline, we enable the screening of the local model storage, differencedetection and the interactive visualisation of the results.
PK-DB 1.0: a pharmacokinetics database for individualizedand stratified modeling Poster 8
PostersessionsFoyer
Jan Grzegorzewski1, Matthias Konig1
1Humboldt University Berlin (Germany)
see page 58
EnzymeML - a SBML-based exchange format forbiocatalytic data Poster 9
PostersessionsFoyer
Colin Halupczok1, Jurgen Pleiss1
1Stuttgart University (Germany)
EnzymeML has been designed as an SBML-based exchange format for data onenzyme-catalyzed reactions. It integrates experimental data such as time course ofsubstrates or products and metadata on the enzyme and the reaction conditionswith modelling data such as the kinetic model and the kinetic parameters obtainedby fitting of the time course data by the kinetic model. Thus, EnzymeML makesdata and metadata related to a biocatalytic experiment findable, accessible, inter-operable, and reusable (Wilkinson 2016). Most of the biocatalytic data is describedby SBML (Hucka et al., 2018) and is enriched by MIRIAM (Novre et al., 2005) an-notation and by additional EnzymeML annotations, the SBML file is included intoa Combine Archive, which includes additionally the original data on which basis themodeling is made. In EnzymeML the structure of the experiment and the modelingpart are separated in different SBML files. This allows the reuse of the same data fordifferent modeling approaches to identify the parameters of the model. EnzymeMLis designed to help the experimenters with a standardized file format, which is based
65
on current standards to easily write, model and publish results of biocatalytic ex-periments. While experimenters can parse their manual designed experiment tablesinto EnzymeML, the container file can be used to model the experiment directly ina modeling platform with the attached experiment data. All the results can be pub-lished then on different platforms, because EnzymeML contains the most relevantdata which is needed by other scientist to do their research.Hucka, M. et al (2018). The Systems Biology Markup Language (SBML): LanguageSpecification for Level 3 Version 2 Core. Journal of Integrative Bioinformatics, 15(1).https://doi.org/10.1515/jib-2017-0081
Novre, N. Le et al (2005). Minimum information requested in the annotation ofbiochemical models (MIRIAM). Nature Biotechnology, 23(12), 15091515. https:
//doi.org/10.1038/nbt1156
Wilkinson, M. D. et al (2016). The FAIR Guiding Principles for scientific datamanagement and stewardship. Scientific Data, 3(1), 160018. https://doi.org/
10.1038/sdata.2016.18.
The JSBML project: a fully featured Java API for workingwith systems biology modelsPoster 10
Postersessions
FoyerThomas M. Hamm1, Nicolas Rodriguez2, Roman Schulte1, Leandro Watanabe3,Ibrahim Y. Vazirabad4, Victor Kofia5, Chris J. Myers3, Akira Funahashi6, Michael
Hucka7, Andreas Drager1,8
1Univerisity of Tubingen (Germany), 2The Babraham Institute, Cambridge, (UK),3University of Utah (USA), 4Marquette University, Milwaukee (USA), 5University ofToronto (Canada), 6Keio University (Japan), 7The California Institute of Technology
(USA), 8German Center for Infection Research (DZIF) (Germany)
Background: SBML is the most widely used data format to encode and exchangemodels in systems biology. The open-source JSBML project has been launched in2009 as an international collaboration with the aim to provide a feature-rich pureJava implementation for reading, manipulating and writing SBML files.Results: The JSBML project has matured into a stable, actively developed, andwell-documented software project with a large number of contributors around theworld. A growing number of applications is now available that uses JSBML astheir back-end for data manipulation. These cover diverse areas of use cases, suchas model building and graphical display, constraint-based modeling, dynamic sim-ulation, model annotation, and many more. JSBML supports all levels, versions,and releases of SBML and provides numerous utility functions that facilitate work-ing with this standard. JSBML also integrates well with other Java libraries forcommunity standards, such as for SBGN or the COMBINE Archive format.Discussion: The JSBML team actively maintains and updates the project. JSBMLis being used in students education and numerous research projects. Major modeldatabases, such as BioModels or BiGG Models, use JSBML-based tools for theircuration pipelines. JSBML is also regularly subject of international students codingevents.
66
Availability: Source code, binaries and documentation for JSBML can be freelyobtained under the terms of the LGPL 2.1 from the website http://sbml.org/
Software/JSBML/ and on GitHub https://github.com/sbmlteam/jsbml/. Theusers guide at http://sbml.org/Software/JSBML/docs/ provides further informa-tion about using JSBML.References:Drager, A., et al (2011). JSBML: a flexible Java library for working with SBML.Bioinformatics, doi:10.1093/bioinformatics/btv341.Rodriguez, N., et al (2015). JSBML 1.0: providing a smorgasbord of options to en-code systems biology models. Bioinformatics, doi:10.1093/bioinformatics/btr361.
Identifiers.org Compact Identifiers for robust data citation Poster 11PostersessionsFoyer
Henning Hermjakob1, Sarala Wimalaratne1, Manuel Bernal Llinares1, JavierFerrer1, Nick Juty2
1European Bioinformatics Institute (EMBL-EBI) (UK), 2University of Manchester (UK)
see page 42
Experiences with Developing Digital Registries forContraceptive Care using the Collaborative Requirements
Development Methodology Poster 12PostersessionsFoyer
Esther Inau1
1University Medicine Greifswald (Germany)
Digital registries linked to standardised medical terminology are increasingly be-coming an entry point to adopting patient health records and facilitating popu-lation level tracking of health services. They support interoperability and offer adiversity of functionalities including vital events tracking and electronic decisionsupport. However, these systems can only be appropriately designed by first gettinga comprehensive picture of the actual tasks and processes associated with serviceprovision; the issues and barriers experienced; and the content and flow of data -all within the local context and aligned with the local infrastructure. This studyexplores the collaborative requirements development methodology to analyse com-mon workflows of clinicians working in contraceptive care and identify functionaluser requirements that facilitate development of supportive digital registries. Theresults of this study can be customized and used as guidance for software profession-als developing national digital registries for use in various healthcare settings andcontexts worldwide. Keywords: digital registries, medical terminology, workflows,functional user requirements, contraceptive care.
67
Virtual patientPoster 13Poster
sessionsFoyer
Fedor Kolpakov1
1Institute of Computational Technologies SB RAS, (Russia)
We have developed a technology for constructing a virtual patient and optimizingthe choice of drug therapy using the example of treating arterial hypertension. Toachieve this goal we have solved three main tasks:
1) building a modular mathematical model of human biochemistry and physiol-ogy with a sufficient level of detail for a given disease. We believe that now it is notrealistic to build a ”virtual patient” for all occasions. Therefore, our approach is tocreate a set of basic blocks, and from them to assemble a model for a given patientand illness (as from Lego blocks).
2) for the main classes of antihypertensive drugs: direct inhibitors of renin(aliskiren), calcium channel blockers (amlodipine), angiotensin II receptor antag-onists (losartan, azilsartan), angiotensin-converting enzyme inhibitors (enalapril,perindopril, lysinopril) and thiazide-like diuretics, their points of impact on the con-structed model were determined and corresponding models of pharmacokinetics andpharmacodynamics were constructed. To validate the resulting model, we used datafrom clinical studies found in the literature.
3) for model personalization we used data from medical records. However thesedata are insufficient to estimate values for many model parameters. In order tosolve the problem a multitude of virtual patients was built where the unknownparameter values can vary significantly. After that we simulated the effects of theabove antihypertensive drugs. Each virtual patient responds to the treatment inhis own way, it will not be effective for everyone. Further we can select groups ofvirtual patients with a similar reaction to the drug and determine which parametersdetermine the division into these groups.
Using the proposed approach, the treatment of arterial hypertension was modeledfor six real patients. The proposed algorithm provides a fairly accurate predictionof the treatment of the selected antihypertensive drug for a particular patient.
As a practical result of the work, a computer program was developed. It worksas follows: the available patient data is entered and the program creates manyvirtual patients for whom it predicts the most likely effect of their treatment andwhich unknown personal parameters need to be determined in order to make a moreaccurate choice. However, to implement this program in medical practice, there isstill a long way to go - to test it on a large number of patients and go through thecertification process.
68
FAIRDOM approach for Systems Biology Data and ModelManagement Poster 14
PostersessionsFoyer
Olga Krebs1, Martin Golebiewski1, Stuart Owen2, Maja Rey1, Natalie Stanford2,Ulrike Wittig1, Xiaoming Hu1, Katy Wolstencroft3, Jacky L. Snoep2,4, Bernd
Rinn5, Wolfgang Muller1, Carole Goble2
1Heidelberg Institute for Theoretical Studies (HITS) (Germany), 2University ofManchester (UK), 3Leiden University (NL), 4Stellenbosch University (South Africa),
5ETH Zurich (Switzerland)
Systems Biologists need a data management infrastructure that enables collabo-rating researchers to share and exchange information and data as and when itis produced, throughout the entire iterative cycle of experimentation and mod-elling. We develop and offer integrated data management support for systemsbiology research within and across research consortia comprising a whole pack-age of solutions. This is applied to large-scale research initiatives in which weare responsible for the scientific data management, like the German Virtual LiverNetwork (http://www.virtual-liver.de/), MESI-STRAT (Systems Medicine ofMetabolic-Signaling Networks https://mesi-strat.eu/) , and European researchnetworks like ERASysAPP (ERA-Net for Systems Biology Applications), SysMO(Systems Biology of Microorganisms) or NMTrypI (New Medicines for Trypanoso-matidic Infections), and Synthetic Biology Centres at Manchester (SynBioChem)and Edinburgh (SynthSys). Our data management concept consists of 4 major pil-lars: 1) Infrastructure backbone: The SEEK platform as registry and a commonsfor data, models, processes and resulting publications and presentations, at thesame time yellow pages for projects, people and events 2) Terminology: Tailoreduse of controlled vocabularies and ontologies to describe the data 3) Modelling sup-port: Seamless handling and simulation of models by integrated modelling platforms(JWS-Online, SYCAMORE, Cytoscape) 4) Social support: Data management ad-vocates within the projects for gathering requirements and dissemination The datamanagement concept that we have developed is not only applied in the research con-sortia that we are responsible for, but also used by other systems biology projects.
NESTML: A domain-specific language for biologicallyrealistic neuron and synapse models Poster 15
PostersessionsFoyer
Charl Linssen 1, Jochen M. Eppler1, Abigail Morrison2
1Forschungszentrum Julich GmbH (Germany), 2Institute of Cognitive Neuroscience,Ruhr-University Bochum (Germany)
see page 55
69
Physiome - Publish your models curated for reproducibilityand reusability to increase research quality for everyonePoster 16
Postersessions
Foyer
Karin Lundengard1, Peter Hunter1, David Nickerson1
1Auckland Bioengineering Institute, University of Auckland (New Zealand)
see page 32
Modelling pancreas - Computational modelling and analysisof altered metabolism in the pancreas and its relationship
to diseasePoster 17Poster
sessionsFoyer
Deepa Maheshvare1 Soumyendu Raha1, Matthias Konig2, Debnath Pal1
1Indian Institute of Science (India), 2Humboldt-University Berlin (Germany)
see page 57
Enhancing the computational based function annotation ofthiamine diphosphate dependent enzymesPoster 18
Postersessions
Foyer
Stephan Malzacher1, Patrick Buchholz, Colin Halupczok, Saskia Bock, DorteRother, Jurgen Pleiss
1Forschungszentrum Julich (Germany)
Thiamine diphosphate (ThDP)-dependent enzymes are potential stereoselective cat-alysts for the synthesis of pharmaceutical precursors like (S)-phenylacetylcarbinol[1]. Most importantly, the process design for the biocatalytic synthesis of these pre-cursors necessitates the identification of suitable ThDP dependent enzymes. There-fore, public database systems provide a large amount of protein sequences, with afunctional annotation by either sequence or structure based homology modelling [2].However, this functional annotation is error prone. Homology models are createdand evaluated by the comparison of experimental data with the protein sequencesand structures of already characterised enzymes. These data are generally obtainedfrom peer-reviewed articles. One approach to enhance the accuracy of computa-tional function annotation is to increase the number of protein sequences linked toexperimental results. Therefore, the collection of experimental data, optionally di-rectly from the experiments conducted in the laboratories, is desired. Linking thesedata to information like the formulation of the catalyst, the reaction environmentand the analytical methods, enable the reproducibility, comparison and evaluation ofthe gained results. For this purpose, the thiamine diphosphate-dependent enzymeengineering database (TEED) [3, 4] enables the opportunity to directly transmitdata gained from experiments with ThDP-dependent enzymes. Thus, the TEEDgrants a way for achieving and further process biocatalytic data in a standardisedway, enabling reproducibility. This approach not only increases the number of data
70
linked to ThDP-dependent enzyme sequences, which can be further applied for mod-elling. Moreover, the negative results of reactions, which are usually not published,can be gathered as an important source for function modelling.[1] Sehl, T. et al (2017) Green Chem. 19, 380384. DOI: 10.1039/C6GC01803C.[2] Copp, J. N. et al Akiva, E., Babbitt, P. C., and Tokuriki, N. (2018) Biochemistry57, 46514662. DOI: 10.1021/acs.biochem.8b00473.[3] Buchholz, P. C. F. et al (2016) ChemBioChem 17, 20932098.[4] Vogel, C., and Pleiss, J. (2014) Proteins 82, 25232537. DOI: 10.1002/prot.24615.
Visualization of Part Use in SynBioHub Poster 19PostersessionsFoyer
Jeanet Mante1, Zach Zundel1, Chris Myers1
1University of Utah (USA)
see paper 37
Comprehensive Modelling Platform Poster 20PostersessionsFoyer
Lukrecia Mertova1, Matej Trojak, Jan Cerveny, Marek Havlı, Matej Hajna,Radoslav Doktor, Ondrej Lostak
1Masaryk University, Faculty of Informatics (Czech Republic)
Comprehensive Modelling Platform (CMP) combines model building, model analy-sis, and annotation tasks in a single public site related to a system of interest. Modelanalysis module is designed to ensure the compatibility with SED-ML to provideexchangeability of simulation experiments.The models are available in the modelrepository and they are in compliance with SBML level 3. The key feature of the in-stantiation for a given system relies on mapping kinetic models to Biochemical Space- a precise representation of the related biological knowledge, thereby supporting thesystems biological view of the modelled system. Besides the model repository, theplatform includes experiments repository and relevant Bioquantities, which improveexperimental verification of the models. The general goal of the platform is to re-spect the need for maintaining existing ODE models but allows to align them witha mechanistic rule-based description that is understandable by biologists, compactin size, executable in terms of allowing basic analysis tasks ensuring consistency ofthe description, and provides links to existing bioinformatics annotation databases.Such a comprehensive solution allows supporting the effort of modellers in buildingmathematical models that have clear biochemical meaning and can be easily inte-grated. Moreover, the mechanistic description can be later used as computationalmodels having all advantages of rule-based modelling. In contrast to existing toolssuch as Biomodels, CellML, or JWSonline which provide general repositories for bi-ological models, the platform is directly focused on a single system. Moreover, noneof these tools is coupled with a systematically organized biological background.
71
PySB framework: Tools to build, calibrate and visualizebiochemical modelsPoster 21
Postersessions
FoyerOscar O. Ortega1, Carlos F. Lopez1
1Vanderbilt University (USA)
see page 38
The Center for Reproducible Biomedical ModelingPoster 22Poster
sessionsFoyer
Anand Rampadarath1, David Nickerson1, Michael Blinov2, John Gennari3,Arthur Goldberg4, Jonathan Karr4, Herbert Sauro3
1Auckland Bioengineering Institute, University of Auckland (New Zealand), 2Universityof Connecticut School of Medicine (USA), 3University of Washington (USA), 4Icahn
School of Medicine at Mount Sinai, New York (USA)
The Center for Reproducible Biomedical Modeling is a NIH-funded center whichaims to enable comprehensive predictive models of biological systems, such as whole-cell models, that can guide medicine and bioengineering. Achieving this goal requiresnew tools, resources, and best practices for systematically, scalably, and collabora-tively building, simulating and applying models, as well as new researchers trainedin comprehensive modeling. To meet these needs, the center is developing newtechnologies for comprehensive modeling, working with journals to provide authors,reviewers, and editors model annotation and validation services, and organizingcourses and meetings to train researchers to model systematically, scalably, andcollaboratively. For more information see: http://reproduciblebiomodels.org.
Datanator: Tools for Aggregating Data for Large-ScaleBiomodelingPoster 23
Postersessions
FoyerYosef Roth1, Zhouyang Lian1, Saahith Pochiraju1, Jonathan Karr1
1Icahn School of Medicine at Mount Sinai (USA)
see page 36
72
Standardizing Electronic Laboratory Notebooks to improvetheir integration with data from the biomedical domain Poster 24
PostersessionsFoyer
Max Schroder1, Dagmar Waltemath2, Frank Kruger1, Sascha Spors1
1University of Rostock (Germany), 2University Medicine Greifswald (Germany)
The FAIR principles [2] require a semantic documentation of research data. Re-search datasets following these principles are beneficial for the scientific communityas they can be semantically interconnected with other datasets and, thus, be reused.Semantic interoperability, however, requires metadata standards to be employedthroughout the entire research process. In the biomedical domain, many experi-ments are conducted in the wet-lab and documented by laboratory notebooks. Therecent shift from paper-based laboratory notebooks to electronic laboratory note-books (ELNs [1]) enables researchers to build their documentation upon templatesto enforce a standardised documentation.
In this poster, we advocate the use of ELNs together with Minimum Informationguidelines and standards for laboratory data. The early adoption of standardizedvocabularies, taxonomies, and ontologies, such as UMLS, MeSH, or ICD-10 willsupport the uptake of ELNs and ease the documentation process. The applicationof standards during the documentation process will in turn ease the creation of FAIRresearch data, and contribute to the further development of automated mechanismsfor data extraction from ELNs.
As a result, research data will gain higher quality and be collected following morestandardized operating procedures, thus, fostering the reusability of work done inthe laboratory. Beside the standardization of ELNs in the biomedical domain, theclinical domain will benefit from the possibility to create structured, semanticallyenriched knowledge that could be leveraged in electronic health records.[1] S. Y. Nussbeck et al. The laboratory notebook in the 21st century: The electroniclaboratory notebook would enhance good scientific practice and increase researchproductivity. EMBO reports, may 2014.[2] Mark D. Wilkinson et al. The FAIR guiding principles for scientific data man-agement and stewardship. Scientific Data, 3:160018, mar 2016.
Minibatch optimization: a method for training mechanisticmodels using large data sets Poster 25
PostersessionsFoyer
Paul Stapor1, Leonard Schmiester, Jan Hasenauer, Daniel Weindl1Institute of Computational Biology, Helmholtz Zentrum Munchen (Germany)
see page 57
73
A Virtual Cohort of Heart Failure Patients Four-chamberHeart Meshes for Cardiac Electro-mechanics SimulationsPoster 26
Postersessions
Foyer
Marina Strocchi1, Christoph Augustin2, Matthias Gsell2, Orod Razeghi1, AntonPrassl1, Edward Vigmond3, Jonathan Behar4, Justin Gould4, Sidhu Baldeep4, Aldo
Rinaldi4, Martin Bishop1, Gernot Plank2, Steven Niederer1
1King’s College London (UK), 2Medical University of Graz (Austria), 3University ofBordeaux and LIRYC Electrophysiology and Heart Modeling Institute (France), 4King’s
College London and Guy’s and St Thomas’ NHS Foundation Trust (UK)
see page 55
SABIO-RK: kinetic data for systems biologyPoster 27Poster
sessionsFoyer
Andreas Weidemann1, Ulrike Wittig1, Maja Rey1, Wolfgang Muller1
1Heidelberg Institute for Theoretical Studies (HITS) (Germany)
The SABIO-RK database (http://sabiork.h-its.org/) supports modelers of bio-chemical reactions and complex networks and is part of the German bioinformaticsnetwork (de.NBI). SABIO-RK represents a repository for structured, curated, andannotated data on reactions and their kinetics. The data are manually extractedfrom the scientific literature and stored in a relational database. The content com-prises both naturally occurring and alternatively measured biochemical reactions,and the data are made available to the public via a web-based search interface. Ad-ditionally, web services can be used to automatically access the database, which isalso used for the retrieval of kinetics data by third-party software tools and dataworkflows. These tools include CellDesigner, VirtualCell, Sycamore, SBMLsqueezer,cy3sabiork, Path2Models, LigDig, and FAIRDOMHub. Data are highly interlinkedto external databases, ontologies, and controlled vocabularies. Recent work includesthe linking to Biomodels (based on Chebi and UniProt identifiers), Rhea, Reactome,MetaCyc and MetaNetX (based on the corresponding reaction identifiers). Addition-ally, Schema.org semantic markup was added to web pages to improve SABIO-RKdata interoperability in life sciences.
Flapjack: an open-source tool for storing, visualising,analysing and modelling kinetic gene expression dataPoster 28
Postersessions
Foyer
Guillermo Yanez1, Isaac Nunez1, Tamara Matute1, Fernan Federici1, TimothyRudge1
1Pontificia Universidad Catolica de Chile (Chile)
see page 43
74
Standardised meta-data in Hospital Information Systemsunlock clinical data for research purposes A real-world
example Poster 29PostersessionsFoyer
Robert Gott1, Kai Fitzer1, Torsten Leddig1, Dagmar Waltemath1, ThomasBahls1
1University Medicine Greifswald (Germany)
see page 59
75
76